Episode Summary

In this month’s AI Double Take, LawDroid CEO Tom Martin and Chief Legal Futurist Sateesh Nori open with a milestone: the launch of LawDroid’s Legal Aid Plugin, an open-source tool built on Anthropic’s Claude platform that fills the legal aid gap Anthropic’s own legal plugins left entirely unaddressed. From there, the hosts unpack a genuinely startling study in which AI responses were preferred over human professors’ answers by law professors themselves (76% of the time, blind test) and discuss a companion research paper showing AI actually helps law students learn the law better. They close with big-picture market moves: OpenAI stepping firmly into legal AI, Anthropic filing for its IPO, and Kirkland & Ellis committing $500 million to AI investment. The thread running through it all: the pieces are finally coming together, and the summer of legal tech is here.

Key Takeaways

1. LawDroid’s Legal Aid Plugin — Filling the Gap Anthropic Left

Just three weeks after Anthropic released its Claude legal plugins, LawDroid launched the Legal Aid Plugin: open-source, built on Claude, and targeted specifically at legal aid organisations and civil legal access. Anthropic’s plugins addressed the commercial legal market; the legal aid segment was entirely absent. Tom credited Anthropic’s architecture, skills, plugins, and the Claude platform, while noting that LawDroid’s decade of access to justice focus made it the natural team to build in that corner of the sandbox. Early feedback highlighted strong interest alongside privacy and security concerns, which the team is actively addressing.

2. From Reductive to Expansive — AI as a Productivity Shift

Sateesh drew a sharp distinction between two modes of AI use. Reductive use: summarising, translating, comparing, going faster through existing work. Expansive use: generating something new that takes you two or three steps ahead, a document, a workflow output, a legal instrument. The Legal Aid Plugin marks a move from the first mode to the second. The shift matters because it changes what is actually possible, not just how quickly existing things get done. Tom added: it’s about better, not just more.

3. The Institutional Knowledge Problem — And AI’s Answer

Tom identified one of the most underappreciated problems in legal aid: institutional knowledge that lives in practitioners’ heads walking out the door when they leave. The Legal Aid Plugin and its underlying skills architecture is a mechanism for capturing that knowledge, codifying it, and making it available to everyone who comes after: customisable, shareable, and open-source. The old paradigm made organisations dependent on vendors. The new one makes practitioners co-authors of the tools themselves.

4. AI Preferred Over Law Professors — By Law Professors

A study using Gemini 2.5 presented law professors with a blind comparison: student questions answered either by a human professor or by AI. The result: law professors preferred the AI-generated responses 76% of the time, without knowing which was which. More striking still: human professors gave answers the study classified as “harmful,” responses that actively hindered student learning, more frequently than the AI did. Sateesh’s takeaway: if AI can fool legal educators who do this for a living, arguments that AI cannot be good enough for everyday people no longer hold.

5. AI Helps Law Students Learn Law Better

A companion study by Professor Daniel Schwartz at the University of Minnesota Law School set out to test the common-sense hypothesis that AI use would hurt law students’ understanding. The result was the opposite: students who used AI learned the law better. The reason: AI provided the scaffolding to understand what the rule actually was something law school casebooks deliberately obscure through Socratic hazing. Tom described it as the ultimate study guide. Sateesh’s implication: AI courses should be introduced on day one of law school, not in the second or third year, and UC Berkeley Law’s current restrictive policy likely runs counter to both studies’ findings.

6. Richard Susskind’s Outcome Thinking Applied

Tom invoked Richard Susskind’s concept of “outcome thinking” to reframe the AI-in-education debate. If the outcome law school exists to produce is students who understand the law and succeed in legal careers, then the question is not who or what delivers the instruction; it is whether the outcome is achieved. Both studies say AI improves the outcome. Process-centered arguments against AI adoption, whether in legal education or legal services, do not survive contact with outcome-focused analysis.

7. The UPL Parallel — Scarcity as System Design

Sateesh connected the law school access debate to the broader unauthorized practice of law problem: both are systems designed around scarcity, limiting who can deliver legal knowledge regardless of quality. The cost of maintaining law schools as gatekeepers (in tuition, time, and the legal labour market they create) is borne disproportionately by people who cannot access legal help at all. AI breaks the scarcity assumption. The model is broken, and the studies provide evidence for why it needs to change.

8. OpenAI Enters Legal AI — Both Giants Now in the Market

OpenAI has followed Anthropic in making a clear, committed move into legal AI. With both leading foundation model companies now focused on the legal market, Tom sees the concentration of capital, talent, and technology as ultimately beneficial for clients, even if the immediate motivation, as Sateesh quipped, may partly be to reduce their own legal fees before their IPOs.

9. Anthropic Files for IPO — and Kirkland Commits $500M

Anthropic has filed for its IPO. Separately, Kirkland & Ellis, the highest-grossing law firm in the world at approximately $10 billion in annual revenue, has committed $500 million to AI investment. Tom’s read: as a fraction of revenue it is not betting the farm, but it is a meaningful signal. The lesson for smaller firms is not to match the dollar amount but to adopt the underlying habit: consistent commitment to innovation, collaboration, and communication.

Show Notes

Topics Covered

• LawDroid Legal Aid Plugin launch — open-source, built on Claude, targeting legal aid market

• Anthropic’s Claude legal plugins and the legal aid gap they left unaddressed

• Privacy and security concerns raised during the plugin launch webinar

• Institutional knowledge capture: skills as a mechanism for preserving practitioner expertise

• The shift from reductive AI use (efficiency) to expansive AI use (productivity)

• AI co-authorship of tools vs. vendor dependency in the old software paradigm

• Blind study: law professors preferred AI responses (Gemini 2.5) 76% of the time

• Human professors gave “harmful” answers more often than AI in the same study

• Professor Daniel Schwartz (University of Minnesota Law School): AI helps law students learn law better

• Law school AI courses currently offered in year 2–3; argument for day-one integration

• UC Berkeley Law’s restrictive AI policy critiqued in light of both studies

• Richard Susskind’s “outcome thinking” as a framework for evaluating AI in legal education

• UPL parallel: scarcity-by-design in law schools mirrors scarcity-by-design in legal services

• OpenAI entering legal AI — both major foundation model companies now in the market

• Sateesh’s theory: AI legal tools partly motivated by reducing IPO legal costs

• Anthropic IPO filing

• Kirkland & Ellis $500M AI investment commitment

• The habit of innovation as the takeaway for small and mid-size firms

People & Organizations Mentioned

• Tom Martin — CEO & Founder, LawDroid

• Sateesh Nori — Chief Legal Futurist, LawDroid

• Professor Daniel Schwartz — University of Minnesota Law School; co-author of AI/law student learning study

• Richard Susskind — Legal futurist; “outcome thinking” concept cited

• Anthropic — Claude platform; released Claude legal plugins; filed for IPO

• OpenAI — Entered legal AI market

• Kirkland & Ellis — Committed $500M to AI investment; ~$10B annual revenue

• Gemini 2.5 (Google) — AI model used in the blind law professor preference study

• UC Berkeley Law — New AI policy noted as running counter to the research findings

• LawDroid — Released Legal Aid Plugin (open-source, Apache 2.0) on Claude platform

Final Takes

Sateesh Nori:

“The next two or three months are going to be huge. I can’t even imagine what’s to come — the OpenAI legal rollout, what Anthropic does next, the impact of the IPO on big legal tech companies. The summer is going to be the summer of legal tech.”

Tom Martin:

“Things are finally coming together. Kirkland’s $500 million is a big headline, but they make $10 billion a year — it’s a good start, not betting the farm. And you don’t need $500 million to do this. You just need the habit of innovation, collaboration, and communication. If you build that habit, you’ll do well.”

AI Double Take is produced by LawDroid | lawdroid.com

If you want to understand how human-centered design and lived experience can transform AI into a genuine access-to-justice engine, you need to listen to this episode. Claire is at the forefront of legal tech product development and brings a uniquely grounded perspective, rooted in years of working directly with the people she now builds for.

Encoding Empathy into AI: How Upsolve Is Rewriting Access to Justice

Join me as I interview Claire Robinson, Chief Product Officer at Upsolve.

In this podcast episode, Claire shares her remarkable journey from growing up in the Redwood forests of Humboldt County to becoming the product mind behind one of the most impactful legal AI tools in the country. She dives deep into how her time as a paralegal, sitting across from low-income Americans navigating bankruptcy, gave her the exact knowledge needed to design an AI tool that truly helps rather than just answers questions. Claire also walks us through how Upsolve’s AI Paralegal works: a context-aware assistant embedded directly inside the app that can see what users see and guide them through the process in real time.

Her stories and insights underscore the power of starting with people, not technology. This episode is a must-watch for anyone curious about how access-to-justice organizations are harnessing AI to do more for the people who need it most, and what it takes to build legal tech that actually works at scale.

The Skinny

Claire Robinson, Chief Product Officer at Upsolve, brings a rare combination of frontline legal experience and product leadership to one of the most important access-to-justice platforms in the country. Upsolve functions as a TurboTax-style tool for Chapter 7 bankruptcy, and has helped over 23,000 families discharge more than $1.1 billion in debt, mostly medical bills and past-due credit cards. Claire began her career at Upsolve as a paralegal, spending thousands of hours helping low-income Americans navigate the bankruptcy process, before transitioning into product leadership and ultimately becoming CPO. That lived experience is the foundation of Upsolve’s AI Paralegal: a context-aware assistant embedded inside the app that can see the same screen as the user, answer questions about the interface in real time, and guide people through a complex legal process on a cell phone, often at night, while managing their families. The tool proved so effective that in November 2024, Upsolve had to shut off new signups for two weeks because demand overwhelmed their human paralegal quality-check team. Claire’s approach, treating AI as an accelerant of good human-centered design, not a replacement for it, offers a powerful model for the broader legal tech world.

Key Takeaways:

• Claire’s origin story, growing up in rural Humboldt County, working in the Bay Area Justice Corps, and helping wildfire survivors, shaped a deep commitment to access-to-justice work that has driven every career decision since

• Her time as a paralegal at Upsolve gave her thousands of conversations with real users, which became the foundation for designing an AI tool that anticipates actual user needs and friction points

• Upsolve’s AI Paralegal goes far beyond a standard chatbot: it is context-aware, meaning it can see what screen the user is on and answer questions about what they are actually experiencing in the app

• The tool proved so effective that Upsolve had to temporarily pause new signups in November 2024, an unprecedented move, because so many people were completing the process that the human paralegal quality-check team couldn’t keep pace

• Upsolve has helped over 23,000 families discharge more than $1.1 billion in debt, and views bankruptcy as the social safety net of last resort for low-income Americans

• Claire’s philosophy is that AI is an accelerant, not inherently good or bad, meaning it amplifies whatever design choices are already baked into the product

• She deliberately took a detour into startup sales at Ditto before joining Upsolve, which gave her breadth of experience that now shows up in her product thinking and cross-functional leadership

• Work-life balance, in Claire’s view, is best measured across a year rather than a week; she structures her life in sprints and finds that doing meaningful work generates more energy, not less

Notable Quotes:

1. “I’m a big fan of the thinking that AI is an accelerant, but not necessarily always a positive accelerant. Just whatever you’re doing, it’s going to be more of that.” - Claire Robinson (05:53-05:59)

2. “So many people were able to get through as a result of the AI chatbot that we had in app that we did actually have to turn off the app for two weeks to any new signups, which was unprecedented in Upsolve history.” - Claire Robinson (02:29-02:41)

3. “We have helped just over 23,000 families relieve, I think now 1.1 billion in debt, mostly medical bills, past due credit cards for basic living expenses.” - Claire Robinson (03:08-03:22)

4. “I wanted to think about what was this tool uniquely positioned to do? And how could we use this as an opportunity to not just provide help, but provide really good help?” - Claire Robinson (27:31-27:42)

5. “If I was entering my name and I entered my first name, but I didn’t enter my last name and that continue button at the bottom was grayed out... I could ask the AI tool, why can’t I move forward? And it would tell me, you need to put in your last name before you can click continue.” - Claire Robinson (28:53-29:04)

6. “As someone who early in my career had to look every day—we could help 20 people and usually about 60 people would show up—when I started playing around with early versions of AI, it was just revolutionary to have this opportunity to not have to turn people away.” - Claire Robinson (26:28-26:50)

7. “I think we have a moral responsibility to make our tool as good as possible as quickly as possible. So if we are in an unfortunate situation where more people need that resource, we’re available to meet that demand.” - Claire Robinson (32:03-32:11)

8. “I do think that there is such a thing as work life balance, but over the course of the year, not a week.” - Claire Robinson (33:44-33:48)

Clips

AI Overwhelmed the App

Balance Happens in Sprints

How a Wild Childhood Shaped Claire

We Have a Moral Responsibility

Claire Robinson’s story to me is ultimately about what happens when the right person builds the right tool at the right time, and makes sure the technology never loses sight of the human being on the other end of the screen. Her path from paralegal to Chief Product Officer is not a detour; it’s a direct line, because every role she held deepened her understanding of what vulnerable people actually need when they’re facing one of the most stressful events of their lives.

What makes Upsolve’s AI Paralegal stand out in a crowded field of legal chatbots is precisely what Claire insists on: the tool knows where you are, what you’re looking at, and what’s tripping you up, because someone spent years in that chair, answering exactly those questions. That embedded empathy is what turned an AI assistant into something capable of overwhelming its own team with success.

Closing Thoughts

What Claire is doing at Upsolve is exactly the kind of work that gets me excited about where legal technology is heading. She’s not building AI for its own sake, she’s building it because she’s sat across from real people who needed help, and she refused to accept the status quo as the permanent state of affairs.

Imagine: Upsolve had to shut its doors to new users because the AI worked too well. That is a remarkable problem to have. It tells you that when you design from genuine human understanding, when your product team has lived the experience your users are living, the results can scale in ways that outpace your own capacity.

For our Legal Rebels community, Claire’s story is a reminder that the most powerful legal technology doesn’t start with the technology. It starts with the person on the other side of the process. The lawyers and legal innovators who will thrive in the years ahead are the ones who hold onto that insight even as the tools get more sophisticated and the temptation to be dazzled by capability grows stronger.

Upsolve has helped 23,000 families be freed from over $1.1 billion in debt. That number is going to keep climbing, and it will do so because someone cared enough to really understand the problem before they tried to solve it.

Paul Simon sat across from Dick Cavett in April 1970 and was asked an impossible question. How does a song come into being? There is a moment when “Bridge Over Troubled Water” does not exist, and then there is another moment when it does, or when it begins to. What happens in between?

Simon reached behind the couch, picked up his acoustic guitar, and demonstrated. He had an opening melody. Then a Bach chorale slipped in from somewhere in his mind. Then he was stuck. The reason he was feeling stuck, he told Cavett, was that “everywhere I went led me where I didn’t want to be.” So he kept listening to a gospel group called the Swan Silvertones, and one of their lines, “I’ll be a bridge over deep water if you trust in my name,” made its way into his thoughts and the song. He smiled and admitted to the audience: “I guess I stole it, actually.”

Then he said: “That’s how songs happen. They piece themselves together.”

I call this The Songwriter’s Method, and I believe it is the most accurate model we have for what serious work with AI actually looks like.

If this sounds interesting to you, read on...

This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.

A Confession About Stealing

What Simon was confessing on national television was that “Bridge Over Troubled Water” was not invented from nothing. It was stitched. The opening melody was his. The next phrase came from Bach, sitting in his subconscious from years of listening. The hook came from a gospel group he had been wearing out on his turntable. The integration was his. The fragments came from everywhere.

This is the part most people miss about creative mastery.

Originality has rarely meant invention from nothing. It has meant a particular ear for which fragments belong together. Eliot wrote that immature poets borrow and mature poets steal. Bach himself was one of the great recombiners in musical history. The myth of the solitary genius spinning new material out of pure inspiration is a Romantic-era story we have inherited and rarely questioned, and it does not match how the work has ever really gotten done.

Simon’s confession is revealing. The song was alive. People recognized themselves in it within the first chord. It has become a standard. And it was built from pieces that already existed in the world. Whatever it was that made the song into a song was not the rarity of the parts, but the ethereal, unique mix.

The Emotional Center

Here is where the conversation between Simon and Cavett turns, and here is the moment that matters most in relation to understanding AI.

Cavett asks Simon what makes him stuck. Simon answers, “Everywhere I went led me where I didn’t want to be.”

On the surface, Simon is describing a compositional problem. Every musical continuation he tried for the half-finished song felt wrong, led to a dead end. But the line means more than that. It is also a description of his life at that moment. He was a young man at the height of his fame, navigating the dissolution of his partnership with Garfunkel, uncertain where to go next as an artist or as a person. Everywhere he was led was everywhere he did not want to be.

That feeling was the songwriting.

The stuckness in the song was the stuckness he was living. And when the Swan Silvertones line came through, what got built around it answered both at once. A bridge appears, in the song and in the life, when someone offers to lay themselves down so you can cross over the trouble, to the other side. The musical fragments resonated because there was an emotional truth at the center pulling them together.

The Same Practice, with More Material

Now bring this back to the question of what serious work with AI actually looks like.

The fragments are everywhere now. AI has made the supply of available material almost infinite. Pieces of every discipline, every style, every register, every voice, every form of structured information, all of it is one prompt away. The Bach chorales of the world are in your pocket. The gospel singers and the scat soloists are in your pocket. So is every textbook, every research paper, every photograph, every line of code, every recipe, every recorded interview.

This is genuinely new. The abundance is real and it changes what is possible.

But, creative work had always been about taste, judgment, curation — the things you don’t choose. It has been about what belongs together, where they touch, what holds them in the same register. That is the songwriter’s work, and the songwriter’s work cannot be outsourced.

What the songwriter does that the AI cannot do is carry the weight of being inside it — to be shattered by betrayal or elevated by love. What it cannot do is be stuck in a life where everywhere you have been led is somewhere you did not want to be. It cannot live in your skin.

How to Be the Songwriter

If you want AI work that comes alive, the practice is closer to what musicians, storytellers, and philosophers have always done than to anything else. Start with what you are carrying inside you: your fears, your thoughts, your questions, your hopes.

Be wide in your search.

Let the Bach chorale and the gospel group and the overheard whispers all come into the room. Do not police the borders of where good material can come from. The legal scholar reaching for cognitive science or psychology, the physicist reaching for philosophy of mind, the engineer reaching for poetry, the parent reaching for evolutionary biology, all of this is the songwriter’s instinct. The fragments are everywhere if you are willing to let them in.

Then trust the integration.

Simon did not consciously decide to fuse Bach and gospel and his own opening line. He listened to the Swan Silvertones over and over because he loved them, and one day the line surfaced and slipped into the song. The integration is partly conscious, partly subconscious, and the subconscious work depends on having marinated in enough material that the connections can form on their own.

Let the AI extend your perception.

Let it surface ideas and connections you would not have found, make connections you would not have made, offer variations you would not have considered. This is real and valuable. The supply of material the AI can put in front of you is one of the genuinely useful capabilities of the technology. The songwriter’s instinct says yes to this. Take the Bach. Take the gospel. Take it all in. Welcome what arrives.

Hold the center yourself.

The AI cannot do this part for you, and trying to delegate it produces exactly the hollow output we have all learned to recognize. What are you actually carrying inside? What is the trouble the bridge is meant to cross?

This works in both directions. The paths the AI surfaces can redirect what you thought you were exploring. A connection the model makes between two of your ideas exposes the deeper question that was driving both. The center can be discovered and refined in the act of gathering. Simon did not start with bridges; he started with a feeling of being led nowhere good, and the bridge arrived as the answer, to find a way home.

Closing Thoughts

Paul Simon’s confession on Dick Cavett’s couch in April 1970 was not really about songwriting alone. It was about how human beings make anything meaningful. We create around what we are living. The work comes together when the center is real and falls apart when it is not.

AI has not changed this.

AI has changed the abundance of what we can pursue, the speed of iteration, the cost of experimentation, the reach of what is available. But the emotional center remains the human condition. It is what we possess that the machine cannot.

The test for whether anything you make with AI is authentic and alive comes down to one question: Did it spring from deep inside you?

Be the songwriter.

Tom Martin is CEO & Founder of LawDroid, Adjunct Professor at Suffolk University Law School, and Author of the forthcoming AI with Purpose: A Strategic Blueprint for Legal Transformation (Globe Law and Business). He is “The AI Law Professor” and writes his eponymous column for the Thomson Reuters Institute.

If you want to understand how a lawyer dismantled everything traditional about legal practice and rebuilt it on first principles (eliminating the billable hour, embracing behavioral economics, and using AI to serve more people with less), you need to listen to this episode. Beau is at the forefront of reimagining family law and brings a rare combination of philosophical depth, entrepreneurial drive, and genuine empathy for the people he serves.

Rebuilding Family Law From the Ground Up

Join me as I interview Beau Atkins, CEO of Evolve Family Law, Canada’s first national family law firm.

In this insightful podcast episode, Beau shares his remarkable journey from Calgary suburbs and construction work to law school, and ultimately to founding a firm intentionally built to operate nothing like a traditional law practice. He dives deep into how lessons from tile-setting, behavioral economics, and a philosophy background converged to shape Evolve Family Law’s fixed-fee model and client-first approach.

Beau also opens up about using AI, including completing a full course on Claude Cowork, to scale a seven-person team to operate with the capacity of seven hundred. His vision of building a TurboTax equivalent for family law is not abstract: it is grounded in the sobering reality that 80% of people with legal problems never reach a lawyer, and that 92% of those needing civil legal aid don’t receive it.

This episode is a must-watch for anyone who believes legal practice can and should be rebuilt to serve people better, offering a practical and deeply human roadmap for doing exactly that.

The Skinny

Beau Atkins, CEO of Evolve Family Law, traces a path from a middle-income Calgary upbringing, shaped by his parents’ divorce at 14, through construction work, a semester in Mexico, and political science studies, to founding Canada’s first national family law firm in 2020. Fueled by a voracious reading habit rooted in behavioral economics thinkers like Daniel Kahneman, Amos Tversky, and Cass Sunstein, Beau rebuilt his legal practice from the ground up, discarding the billable hour in favor of fixed-fee pricing before AI even entered the conversation. Throughout the episode, Beau connects his personal experience of separation with the empathetic, nudge-based counsel he now offers clients, including the powerful reframe of calling a former spouse “your son’s mom” rather than “your ex.” He also discusses how AI and tools like Claude Cowork are enabling his small team to scale access to family law services for the 80% of people with legal problems who never make it to a lawyer’s office, with a purpose he has distilled simply as: to make life better.

Key Takeaways:

• Beau founded Evolve Family Law in 2020 after nearly leaving law altogether, determined to rebuild a firm on best business practices rather than traditional legal conventions

• His background in construction tile-setting gave him a client-first, fixed-fee pricing philosophy long before AI efficiency made the billable hour untenable

• Behavioral economics, particularly the work of Daniel Kahneman and Amos Tversky, deeply informs how Beau thinks about law, client decision-making, and the limits of the “reasonable person” standard

• Beau applies “nudge” theory from directly to his family law practice, including coaching clients to reframe how they refer to a former spouse

• His parents’ divorce at age 14 shaped his trajectory as a family lawyer and his commitment to helping clients navigate separation with more clarity and less conflict

• AI, including completing a full course on Claude Cowork, is enabling Beau’s seven-person team to operate with the capacity of 700, with a vision of building a TurboTax equivalent for family law

• 80% of people with legal problems don’t go to lawyers; 92% of those needing civil legal aid don’t receive it, closing that gap is central to Beau’s mission

• Finding stillness, prioritizing sleep and exercise, and stacking family time with professional activities are how Beau sustains balance across running a firm, a startup, and fatherhood

• Beau’s guiding purpose, articulated through Viktor Frankl’s Man’s Search for Meaning, is simply to make life better, a mission embedded in everything Evolve Family Law does

Notable Quotes:

1. “I don’t need to leave law. I just need to break it all down to the ground and rebuild it on the foundations of how a good business runs using best business practices.” - Beau Atkins (00:18:20-00:18:26)

2. “If what you’re charging an hour for did not take you an hour, I think it’s criminal.” - Beau Atkins (00:21:13-00:21:21)

3. “Lawyers sell peace of mind or certainty. And when you look at the hourly rate, that’s essentially lawyers shirking that responsibility — the very first opportunity they have to give their clients certainty, they’re giving an hourly rate instead.” - Beau Atkins (00:22:50-00:23:08)

4. “That is not your ex-husband. That is not your ex-wife. That is your son’s mom or your daughter’s dad. It’s a lot harder to use expletives when you’re talking about your son’s mom than it is your ex-wife.” - Beau Atkins (00:27:38-00:27:55)

5. “With seven of us, we will be able to perform like we’re seven hundred.” - Beau Atkins (00:32:13-00:32:17)

6. “80% of people who have a legal problem aren’t going to lawyers. 92% of people needing civil legal aid don’t get access to it. Imagine 92% of hungry people just don’t have access to food.” - Beau Atkins (00:32:56-00:33:24)

7. “I am the least qualified, most expensive counselor you’ll find anywhere.” - Beau Atkins (00:29:50-00:29:55)

8. “I’ve identified that my purpose is to make life better. And so long as what I’m doing is helping me fulfill that purpose, I have energy and excitement to how I approach it.” - Beau Atkins (00:38:15-00:38:42)

Clips

Rebuild Law from First Principles

92% of People Can't Access Legal Assistance

Why Lawyers Should Give Fixed Fees

Stop Calling Them “Ex”

Beau Atkins represents something that is rare in law: a practitioner who looked at the profession’s broken incentive structures, understood them through the lens of economics and psychology, and then actually did something about it. From charging fixed fees before AI forced the conversation, to applying behavioral nudges in divorce proceedings, to building a firm designed from day one for access and scale, his approach is systematic, empathetic, and deeply purposeful.

What resonates most is how Beau has woven together threads that most lawyers keep entirely separate: the personal (his parents’ divorce, his own separation in law school), the intellectual (Kahneman, Frankl, Sunstein), and the practical (tile-setting, business books, AI tools). The result is a lawyer who doesn’t just practice law but genuinely rethinks what the practice of law is for.

Closing Thoughts

I’ve spoken with a lot of legal innovators over the years, and what strikes me about Beau is how grounded his disruption is. He’s not innovating for the sake of a headline. He’s doing it because he came up through a working-class family, watched his parents go through divorce, nearly walked away from law entirely, and then decided to rebuild it into something worthy of the people who need it most.

The fixed-fee model, the behavioral economics framework, the Claude Cowork course, the vision of a TurboTax for family law; these aren’t disconnected ideas. They all flow from a single purpose: to make life better. That’s it. Simple, clear, and enormously difficult to actually execute.

What excites me most about this conversation is how clearly it illustrates that the future of legal practice isn’t just about AI tools or pricing models. It’s about lawyers who are willing to interrogate their own assumptions, learn from outside the legal world, and put clients at the center of everything they build. Beau has done that. And his firm is proof that it works.

For the Legal Rebels in our community: you don’t need a philosophy degree or a background in behavioral economics to take something from this episode. You need the willingness to ask, whether the way you practice law is actually designed to serve your clients, or just to serve you. Beau asked that question. The answer changed everything.

I developed a thesis last August I call the Transformation Triangle. The thesis was straightforward and, at the time, not yet obvious. In a world where AI capabilities commoditize on a quarterly cadence, no single moat survives. Not software. Not expertise. Not credentials. The only defensible position is the integration of three elements at once: tools that scale, expertise, and education that transfers capability. Loops, not moats. I built the argument for professional services, but it applies to any business.

Five months later, in February, I returned to the framework to show that the legal AI leaders were already running it. Harvey at an eight-billion-dollar valuation while building Legal Engineers from Wachtell and Latham, then opening Harvey Academy and partnering with seventeen law schools. EvenUp at two billion with more than a hundred in-house experts reviewing every output. Luminance launching academic certifications across three continents. Spellbook in fifty-plus law schools. The market was converging on the thesis.

In April, I refined the framing one more time. What I had called a triangle, three pillars holding up a roof, turned out to be a dynamic system: three currents in a single flow, each leg feeding the next. Software, Services, Learning, closed into a loop. The loop is the moat. Anything less is a feature waiting to be absorbed.

If this sounds interesting to you, please read on...

This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.

Eight Days That Settled the Argument

On May 4, Anthropic announced a new enterprise AI services company. The founding partners were Blackstone, Hellman & Friedman, and Goldman Sachs, with backing from a consortium of alternative asset managers that included General Atlantic, Apollo, GIC, Leonard Green, and Sequoia. The company’s purpose, in the announcement’s own words, is to embed Applied AI engineers from Anthropic alongside the firm’s engineers, working inside mid-market customer organizations to build production Claude systems tailored to those organizations’ operations. The engagement begins by understanding where Claude can have the biggest impact and ends with systems wired into daily work.

Anthropic’s CFO, Krishna Rao, gave the rationale: “Enterprise demand for Claude is significantly outpacing any single delivery model.” Translation: selling the model is not enough. The customer needs the work delivered, not the tool licensed. Or as Richard Susskind might say, the customer needs to hang a picture frame and doesn’t care if it’s accomplished with a hammer and nail or a drill and screw.

On May 12, eight days later, OpenAI acquired the consulting firm Tomoro and launched the OpenAI Deployment Company. The investor consortium included Bain Capital, SoftBank, Warburg Pincus, Capgemini, McKinsey, and Goldman Sachs. The structure was identical in every meaningful respect. Forward Deployed Engineers, the model Palantir made famous, embedded inside customer organizations to design, build, test, and deploy production systems connecting OpenAI models to the customer’s data, tools, and workflows.

Earlier this spring, ahead of either lab’s move, Sequoia partner Julien Bek published an essay called “Services: The New Software.” It went viral, well past three million views on X, four hundred and fifty thousand impressions on LinkedIn. Bek argued that for every dollar a company spends on software, it spends six on services. The next trillion-dollar company, he wrote, will be a software company masquerading as a services firm. Selling the tool means racing the next version of the model. Selling the work means every model release makes the service faster and cheaper. The autopilot, in his vocabulary, captures the labor budget rather than the software budget, and the labor budget is six times larger.

Three of the most influential actors in the AI economy. Three different starting points. One structural answer. Embedded engineering inside the customer, outcome-priced delivery, model access bundled with services, financial sponsorship from alternative asset managers who understood, before anyone else, that the services profit pool dwarfs the software profit pool.

And the customers are not waiting. On May 27, Kirkland & Ellis, the world's highest-grossing law firm, told the Financial Times it would spend $500 million building its own AI platform, with 180 tech professionals working alongside 250 lawyers, including 100 partners, to capture what chair Jon Ballis called "the collective intelligence of our institution." Outside vendors are building the system alongside Kirkland's own engineers and data scientists, but the firm will own all of it. The platform will run mandates in their entirety, under value-based pricing rather than the billable hour. Software, services, learning, owned by the firm, closed into a loop. The labs are building the structure on the supply side. Kirkland is building it on the demand side.

This is the Transformation Triangle.

How the Match Is Exact

Look at the Anthropic announcement again, but this time through the Transformation Triangle’s three legs.

Software is Claude itself, plus the Applied AI tooling that customizes it. That leg already runs.

Services is the embedded engineering team inside the customer, sitting with clinicians and IT staff, building the workflows. That is the leg the new company creates.

Learning is what makes the whole structure compound. Every engagement generates signal: corrections, edge cases, workflow patterns, vocabulary specific to a physician practice or a community bank. That signal flows back into Anthropic, into the Applied AI engineers, into the next deployment, and eventually into the next training run. The loop closes.

The software gets smarter because the services generate signal that the learning leg captures and returns to the software. The services get faster because the software does more of the work and the learning sharpens the judgment behind it. The learning gets richer because both legs are running.

Now look at OpenAI’s structure. Models plus FDEs plus customer deployments plus signal flowing back. Same three legs. Same loop.

Now look at Bek’s essay. Software companies “masquerading” as services firms. The argument is that the system the labs are now building is the only one that works, and the entry point to building it is wherever a category is already outsourcing services that can be re-delivered with AI underneath.

The vocabulary is different in each case. Anthropic says “enterprise AI services.” OpenAI says “deployment.” Sequoia says “autopilot.” I have been calling the underlying structure the Transformation Triangle. The labels differ because the lineages differ, but the structure is the same.

The Model Alone Is No Longer The Product

May 21st, Greg Brockman, President & Co-founder of OpenAI tweeted, “the model alone is no longer the product.”

The companies that built the foundation models, the most successful tools ever shipped in the history of computing, are the same companies that just announced they cannot survive on tool revenue alone. Anthropic and OpenAI created the technology that commoditizes everything built on top of it.

They watched their own customers, the wrapper companies, get eaten by their own product roadmaps. They have now reached the same conclusion every downstream company is reaching: if you sell the tool, you race the next version of the model. If you sell the work, every model release works for you, because every model release makes the work cheaper and faster to deliver.

The model is not the product. The loop is the product. The labs that invented the commoditization technology are not exempt from its consequences. They are simply earlier to the conclusion.

Eight months ago, in a piece called “Every Law Firm Is Now a Software Company,” I described how this would play out across professional services. Three months ago, I showed how it was playing out across legal AI. One month ago, I named the loop as the only structure that survives. This month, the labs themselves confirmed it.

What This Means If You Are Building Anything

The convergence is the signal. If Anthropic and OpenAI cannot survive on tool revenue alone, no one downstream of them can either. The companies that read the signal early and start running the full loop will compound. The companies that wait for further confirmation will discover that they missed the boat.

Three questions require your attention:

Does your Software get smarter from your Services? If your engineering team ships the same product whether or not your delivery team is engaged with customers, the loop is broken on the Software side. You are not building a loop. You are building a feature factory.

Do your Services get faster and more accurate because of your Software? If your delivery team works the same way they worked before AI, you have adopted a tool, not transformed a business. The loop is broken on the Services side. Your delivery economics will not survive a competitor running the full loop.

Does your Learning go anywhere? If the corrections, the patterns, the edge cases your team encounters stay in their heads, in Slack threads, in the memory of a single senior partner, you have brain drain. The loop doesn’t compound. Every engagement starts the same as the last one. Your competitor, the one running a feedback loop, gets one engagement smarter every week.

The answer to any one of those questions for most businesses today, including most law firms, most consulting practices, most software companies, and most professional service firms of every kind, is “No.”

That is the gap. And the gap is what the new lab-services companies, backed by Blackstone and Bain and Goldman and Sequoia, are about to exploit.

Closing Thoughts

I called the Triangle in August. I tracked its validation in February. I submitted my book manuscript to my publisher wherein I refined it in April. This month, the foundation model labs themselves started building it in public. The thesis is not a thesis anymore.

The model is not the product.

The loop is the moat.

Tom Martin is CEO & Founder of LawDroid, Adjunct Professor at Suffolk University Law School, and Author of the forthcoming AI with Purpose: A Strategic Blueprint for Legal Transformation (Globe Law and Business). He is “The AI Law Professor” and writes his eponymous column for the Thomson Reuters Institute.

If you want to understand what the data actually says about AI’s effect on legal reasoning, and why the conversation around lawyers and AI is long overdue for a reality check grounded in evidence, you need to listen to this episode. Daniel is at the forefront of empirical legal scholarship and brings a rigorously grounded, nuanced perspective to one of the most debated questions in legal tech today.

Does AI Dull a Lawyer’s Thinking? The First Empirical Answer Is Here

Join me as I interview Daniel Schwarcz, professor of law at the University of Minnesota and graduate of Harvard Law School.

In this insightful podcast episode, Daniel shares his journey from growing up in Scarsdale, New York, shaped by two lawyer parents, through Amherst College and Harvard Law School, to becoming one of the country’s leading empirical scholars on AI and the legal profession. He dives deep into his landmark randomized controlled trial of 100 law students, the first study of its kind to produce hard empirical evidence on whether using AI for one part of a legal task undermines a lawyer’s reasoning on the rest.

His stories and insights underscore his commitment to grounding the AI conversation in data rather than intuition, including a result that upended his own hypothesis: students who used AI actually performed better even after the AI was taken away. This episode is a must-watch for anyone who cares about the future of legal education, the development of lawyering skills, and how we build an evidence-based understanding of AI’s real impact on the profession.

The Skinny

Daniel Schwarcz, professor of law at the University of Minnesota and a Harvard Law School graduate, has spent nearly two decades building a scholarly career at the intersection of law, economics, and empirical research. His interest in AI predates the generative AI moment by over a decade, rooted in his early work on how machine learning was reshaping insurance underwriting and discrimination. When ChatGPT arrived, Daniel was uniquely positioned to turn that analytical lens onto one of the legal profession’s most pressing anxieties: does using AI make you a worse lawyer?

In a randomized controlled trial of approximately 100 upper-level law students, Daniel and his co-authors tested whether using AI to synthesize legal materials in one stage of a task would undermine performance in a subsequent stage where AI was no longer available. The hypothesis (intuitive, widely held, and seemingly well-supported by evidence from other domains) turned out to be wrong. Students who used AI in stage one not only outperformed the control group when they had AI, they continued to outperform them even after the AI was removed. The mechanism, Daniel explains, is straightforward: AI helped students develop a more accurate, robust understanding of the law, and that superior understanding carried forward.

The episode is a masterclass in how rigorous empirical thinking can cut through the noise of the AI debate and offer lawyers, educators, and legal innovators something genuinely rare: real evidence.

Key Takeaways:

• The first randomized controlled trial of its kind found that law students who used AI to process legal materials performed better in a subsequent AI-free task—the opposite of the prevailing fear

• The mechanism behind the result is intuitive in retrospect: AI helped students synthesize the law more accurately in stage one, and a better understanding of the law carried forward into stage two

• Daniel is careful to scope his findings precisely: the study does not address long-term skill degradation from repeated AI use, a concern he still takes seriously and wants to see tested

• Cognitive fatigue emerged as a potentially important moderating variable: students using AI at the end of a three-hour experiment appeared to use it less critically, with negative consequences for those who had already produced strong work

• The finding that AI helped weaker performers but may have hurt stronger performers in the revision stage points to the importance of how and when AI is introduced, not just whether it is used

• Daniel’s decade-plus background in AI and insurance, studying how machine learning tools affected underwriting and discrimination, primed him to think rigorously about AI’s real-world consequences long before generative AI arrived

• His academic philosophy is built on grounded, practical relevance: he deliberately chose insurance law because it was important, understudied, and tangibly affected millions of people, the same spirit animates his AI research

• Setting clear personal rules and treating them as binding is Daniel’s approach to work-life balance, a discipline he argues is especially necessary for academics, who lack the external structure of a typical workplace

Notable Quotes:

1. “There was still sort of a lot of hesitation among lawyers, because of this fear that using AI will make us worse lawyers. And I take that fear very seriously, and so it’s something I wanted to empirically investigate.” - Daniel Schwarcz (02:43-02:56)

2. “The folks who had AI, their brain was much less engaged. And so there’s something intuitive to me about that.” - Daniel Schwarcz (04:30-04:51)

3. “Our hypothesis going in was that the folks who used AI in stage one to help them make sense of the materials would perform less well once we took the AI away from them. But we found the opposite result.” - Daniel Schwarcz (25:34-26:32)

4. “The effect really was that AI helped you synthesize the rule, helped you get the rule accurately and see the nuances in stage one. And then even when you no longer had AI and you’re applying it to a new situation, having the rule in an accurate, robust fashion was helpful even when the AI was no longer available.” - Daniel Schwarcz (38:21-38:47)

5. “I think there’s good reason to think that when you’re using AI and you sort of have cognitive fatigue, you’re tired, you’re overwhelmed, maybe then actually the AI is more likely to push out critical thinking than if you’re using AI sort of when you’re fresh and you’re motivated.” - Daniel Schwarcz (34:32-34:49)

6. “I always want to sort of caution against saying we’ve definitively proven AI is not a risk. No. I think though that relative to my views on this before doing the study, I have less concern that using AI necessarily or inevitably dulls your capacity to perform when you’re not using it.” - Daniel Schwarcz (35:16-35:45)

7. “I really try to treat my own internal rules as binding because if you don’t, then there’s no point in setting those rules.” - Daniel Schwarcz (44:12-44:22)

8. “Part of why I like doing practical things is I like that it’s relevant to people. I don’t want to be relevant not just to a small group of ivory tower folks, but I want work that’s relevant to people that people find interesting and helpful in their daily lives.” - Daniel Schwarcz (44:48-45:03)

Clips

RCT Finds Surprising AI Effect

AI Improved Legal Skill Even After Removal

When AI Helps Some and Hurts Others

AI Helps — But Lawyers Worry

Does AI Really Make You A Better Lawyer?

Daniel Schwarcz’s work represents something the legal profession badly needs right now: empirical grounding in a conversation that has been dominated by intuition, fear, and hype in equal measure. His finding that AI use did not undermine, and in fact improved, subsequent AI-free performance challenges the most common argument offered by those skeptical of AI adoption in legal education and practice. But what makes this episode especially valuable is Daniel’s own intellectual honesty about what his study does and does not prove. He is not an AI booster. He is a scientist who followed his data, and whose data happened to push back against a widely held assumption.

What stands out most is the nuance he brings to the question of context. Whether AI helps or hurts legal reasoning may depend on factors like cognitive fatigue, the quality of work already produced, and how critically the tool is used. Those are not simple answers, but they are honest ones, and they point toward the kind of sophisticated, situation-specific thinking that lawyers are actually well equipped to do.

Closing Thoughts

What I love about this conversation with Daniel is that it does what the best scholarship always does: it takes a question seriously enough to actually test it. The fear that AI is quietly hollowing out lawyers’ reasoning skills is real and widespread, and it deserves a real answer. Daniel’s study provides one, even if it is a partial and carefully qualified one. And the answer, at least for now, is more reassuring than most people expected.

For me, this gets to something I’ve believed for a long time: we need to stop making decisions about AI in the legal profession based on speculation and start making them based on evidence. Daniel is doing that work. And the results are more nuanced, more interesting, and ultimately more useful than the hot takes dominating our feeds.

What also struck me is how Daniel embodies the very thing that makes a great legal scholar: intellectual rigor combined with genuine care for practical relevance. He chose insurance law because it mattered to real people. He chose to study AI’s effect on lawyers because that question matters to real people. That through line (grounded, practical, consequential) is exactly what the legal profession needs more of as we navigate this moment.

For our Legal Rebels community, the takeaway is not that AI is risk-free. Daniel himself would push back on that. The takeaway is that the risks are more specific, more contextual, and more manageable than the doom-and-gloom narrative suggests. Use AI when you’re sharp. Verify what it gives you. Understand the law it’s helping you synthesize. Do that, and the evidence suggests you may actually come out ahead.

Link to Daniel’s paper:

Bednar, Nicholas and Cleveland, David R. and Erbsen, Allan and Schwarcz, Daniel, Artificial Intelligence and Human Legal Reasoning (April 05, 2026). Minnesota Legal Studies Research Paper 2026-21, Available at SSRN: https://ssrn.com/abstract=6525800 or http://dx.doi.org/10.2139/ssrn.6525800

The phone rings at 8:47 in the morning, and by 8:47 it is already too late. The intake line was full when the office opened. The voicemail box filled by 8:30. The eviction filings from the weekend are sitting in three piles on the receptionist’s desk. There are six staff attorneys in this office, for a county of 250,000 people, and the math has not worked for a long time.

Today, like every day, the office will help everyone it can. And by the close of business, more people will have called than were helped.

This is the work of civil legal aid.

Not the courtroom drama. The part that happens everyday. The eligibility screen on a caller’s third try to reach a lawyer. The motion drafted in twenty minutes between a hearing and a client meeting. The deadline tracked on a sticky note because the case-management system is overloaded. The transfer memo written on a Friday because a paralegal is leaving on Monday. The funder report assembled in the last week of the quarter, from records that should have been clean three months ago. This is the experience of holding the line, every day, against a problem that is never going to be solved by just working harder.

LSC’s 2022 Justice Gap Study put the number at 92 percent. Ninety-two percent of the substantial civil legal problems of low-income Americans get no help, or not enough help, to make a difference. LSC-funded organizations turn away one of every two requests they receive, not because they want to, but because they cannot do otherwise. The capacity is not there. It has never been there. And every legal aid manager I know has spent their career figuring out how to make less than enough go as far as possible.

Today I want to talk about a new thing that the civil legal aid community can use, and what I think it make possible.

If this sounds interesting to you, read on...

This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.

The Legal Aid Plugin

At the LawDroid AI Conference 2026 last month, our theme was “The Year to Build” and we’ve been hard at work.

This week, LawDroid built and published the Legal Aid Plugin: a free, open-source Claude for Legal plugin built for civil legal aid offices, court self-help programs, and public-interest providers. It is published under the Apache 2.0 licence and available immediately at github.com/lawdroidAI/legal-aid-plugin.

“What matters about this launch is not just the plugin. It’s published openly, free to use, and open to contribution from any legal aid organization, anywhere. The civil legal aid community needs technology built that way.”

— Sally Chaffin, Practice Innovation Manager at Atlanta Legal Aid Society

The plugin scaffolds the daily work of a legal aid office. Fifteen skills, designed around the realities of how civil legal aid actually operates: eligibility screening that is aware of every funder the office reports to (LSC, IOLTA, foundation, state); structured intake; document drafting with funder-restriction flagging that propagates through every stage; case analysis memos; research roadmaps; status updates; routine correspondence; deadline tracking; communications logs; transfer memos for the rolling staff transitions every legal aid office knows; supervision queues; and reporting data extraction for the six funder shapes that dominate the sector.

What makes it different from existing tools is not any one skill. It is that the architecture is built for legal aid, not adapted from BigLaw. Funder rules propagate. Ethical discipline is structural, not optional. Supervision is configurable to the three styles legal aid offices actually use. The verification habits are wired in at the prompt level. None of this can be added on top of a tool built for corporate practice; it has to be there from the start.

It runs locally. Every legal aid office in the country can deploy it for free, today.

The Force Multiplier

The Legal Aid Plugin is not a replacement for lawyers. It is not a chatbot. It is not even, on its own, a productivity tool. What it is, in the language of strategy, is a force multiplier: a capability that increases an organization’s effectiveness beyond its body count. A force multiplier does not change how many people are on the team. It changes what the team can accomplish in the time available.

“The need for civil legal aid has always exceeded our capacity, and the gap is widening. Tools like this plugin let us redirect time from administration to advocacy. That’s a real shift in what our staff can do for clients.”

— Pablo Ramirez, Executive Director, Legal Aid Society of San Bernardino

Civil legal aid has been waiting for a force multiplier of this kind for a long time. The sector has tried many: case-management systems, intake routing platforms, document assembly engines, online self-help tools. Each helped, in pockets. None changed the underlying ratio of capacity to need.

What is different now is the architecture beneath the plugin. Anthropic released a way to build structured, composable, supervised AI capabilities that wrap around the lawyer rather than replacing the lawyer. A plugin is a directory of skills, each one capturing what a competent person in this domain would do next, given this context. The skill carries instructions, references, sequencing logic, and access to relevant tools. It runs in conversation with the lawyer carrying it, shaped by what the lawyer brings to the moment. The lawyer is still the lawyer. The plugin is what lifts the procedural load that has always made the job harder than it needed to be.

This is the leverage point the sector has not had before. Not because no one tried, but because the architecture did not exist to make leverage like this possible.

Externalizing What We Already Know

There is a quieter point worth making, because it matters for how this scales.

For decades, professional expertise has been treated as something that lives in our heads. The senior eligibility specialist who can hear, in the first thirty seconds of a call, whether the matter is an emergency. The managing attorney who can read an intake summary and see the three issues no one else flagged. The clinic director who knows which judges grant which motions and how to ask. These people are the institutional knowledge of their offices, and when they leave, much of that knowledge leaves with them. The next staffer rebuilds it slowly, from cases, over years.

The plugin architecture lets us externalize expertise. But, not all of it. The judgment that makes a senior eligibility specialist senior cannot be put into a skill, and we should not pretend otherwise. But the procedures that surrounds judgment, the sequencing of questions, the funder-source matrix, the conflict pre-check, the urgency triage, the routing logic, can be put into a skill. And once it is in a skill, every staffer the office has, today and a decade from now, can carry it from day one.

For civil legal aid in particular, where staff turnover is real and senior practitioners are precious, this is the difference between a sector that loses institutional knowledge every time a key person leaves and a sector that compounds its expertise over time.

What This Makes Possible

The plugin is the launch. What follows is what I am more interested in, because the plugin is one example of something that can keep happening.

More clients served. When a staff attorney spends less time on procedural overhead, they spend more time on advocacy and counsel. When an intake specialist runs a structured eligibility screen in fifteen minutes rather than thirty, the office gets through more of the line. The justice gap is not closed by any one of these. But the offices that hold the line gain capacity they did not have, in a sector where every additional matter served is a real outcome for a real person.

Faster, more consistent onboarding. New staff arriving in a legal aid office have always faced the same hard ramp: weeks to learn the office’s specific procedures, months to absorb funder rules, longer to learn what to listen for. The plugin compresses the procedural part of that ramp. A new staffer who runs /legal-aid:onboard on day one inherits the office’s configured discipline, practice-area guides, supervision conventions, and verification habits from the start.

Institutional discipline that compounds. Each office that deploys the plugin builds practice-area guides, captures its plausibility ranges, configures its supervision style. Over time, that configuration becomes a record of how the office actually works. Cross-organization learning becomes possible in a way it has not been: an office in North Carolina can see how an office in Georgia structured its housing intake; an office in New Hampshire can borrow a funder-restriction analysis from one in Florida.

A platform layer the sector owns. This is the part that matters most over time. Open source under Apache 2.0 means no vendor lock-in, no licensing fees, no dependency on what BigLaw gets. The civil legal aid sector has, for the first time, an AI platform layer it can build for itself, shape over time, and own collectively. Anthropic published the architecture; the community publishes the skills. The plugin we launched this week is the first piece. There will be more.

The Opportunity

The numbers in the Justice Gap Study are not going to be fixed by any plugin. They will be fixed, when they are fixed, by structural change: more funding, more lawyers, more legal aid programs, more pro bono, more reform of the systems that produce these problems in the first place.

One of the things we’ve learned through the work of the Legal Innovation Lab is that the access to justice gap is not going to be solved through a “business as usual” model.

The demand for civil legal help continues to outpace available resources, especially in rural and underserved communities. At Legal Aid of NC we’ve been exploring how technology can reduce administrative burden and give staff more time to focus on advocacy, problem solving, and client support.

It’s encouraging to see developers thinking intentionally about how AI skills and plugins can support legal services and expand access in practical and responsible ways. It’s even better when legal services programs get to help shape what’s being built.”

— Scheree Gilchrist, Chief Innovation Officer, Legal Aid of North Carolina

With the plugin and tools like it, the same six staff attorneys, in the same county of 250,000 people, can serve more matters, more consistently, with more institutional memory, at a higher standard. That is not closing the justice gap. But, it is multiplying the effect of the people who are working on it.

For an underfunded sector that has held the line for decades, on the will and skill of its people, a force multiplier built for its specific needs is a meaningful thing.

Closing Thoughts

The phone still rings at 8:47 in the morning. The voicemail box still fills. The eviction filings still arrive on Mondays. None of that is going to change because of a plugin.

What can change is what happens after the phone rings. The eligibility screen that catches the funder-restriction question on the first pass. The intake that propagates from the caller’s prior conversation rather than starting cold. The deadline that gets recorded with the funder allocation right, the first time. These are small things. They add up to a different morning.

We are building this for the people who stand for justice.

Come build it with us!

Learn More

This week, LawDroid published the Legal Aid Plugin, a free, open-source Claude for Legal plugin built for civil legal aid offices, court self-help programs, and public-interest providers. Fifteen skills, Apache 2.0, available now.

• Repository

• Documentation

• Webinar and Q&A:

  June 1, 2026 at 10:00am Pacific Time

If you are with a civil legal aid program, court self-help centre, or public-interest practice, we would love to build this with you!

Tom Martin is CEO & Founder of LawDroid, Adjunct Professor at Suffolk University Law School, and Author of the forthcoming AI with Purpose: A Strategic Blueprint for Legal Transformation (Globe Law and Business). He is “The AI Law Professor” and writes his eponymous column for the Thomson Reuters Institute.

If you want to understand the scale of the justice gap in America and how AI and intentional collaboration can help close it, you need to listen to this episode. Ron is at the forefront of the civil legal aid movement and brings a uniquely national perspective on what’s working, what’s not, and what must be built next.

Building a More Accessible and Equitable System of Justice

Join me as I interview Ron Flagg, President of the Legal Services Corporation (LSC).

In this episode, Ron shares his vision for how AI can help address one of the most pressing crises in American law: the justice gap. As the head of the nation’s largest funder of civil legal assistance, Ron brings an unparalleled vantage point: one that spans hundreds of grantee organizations, millions of underserved clients, and the halls of Congress. He dives deep into why funding alone can never close the justice gap, and why the answer lies in the innovators, not just the innovations.

Ron also shares insights from LSC’s landmark eviction report, revealing how the patchwork of local eviction laws across the country produces wildly different outcomes for tenants, and how technology, collaboration, and even landlord partnerships can be part of the solution. This episode is a must-listen for anyone who cares about access to justice, legal innovation, and the role technology can play in building a fairer legal system.

The Skinny

Ron Flagg, President of the Legal Services Corporation, delivers a compelling keynote and fireside chat recorded at the LawDroid AI Conference 2026. With clarity and conviction, Ron lays out why the justice gap — 92% of the civil legal needs of people living in poverty go unmet — cannot be closed by funding or technology alone. He makes the case that it is the community of innovators, not any single innovation, that holds the key. Drawing on LSC’s deep grantee network and its culture of collaboration, Ron describes a field that isn’t hesitating in the face of AI: it’s hungry. He outlines the most promising internal use cases for AI in legal aid, from streamlining intake and triage to reducing administrative burden on staff. Ron also walks through LSC’s recent eviction report, highlighting the enormous variation in local eviction laws, the ripple effects of housing instability, and the surprising economic case for landlords to partner with tenants rather than evict them. He closes with a clear-eyed update on LSC’s funding outlook and a rallying call: collaboration is a multiplier, and this year is a year to build, intentionally, together.

Key Takeaways:

• The justice gap is staggering: 92% of the civil legal needs of people experiencing poverty go unmet, and LSC grantees are forced to turn away roughly half of all eligible applicants due to lack of resources

• No single solution, whether funding or technology, can close the justice gap; the answer lies in a community of innovators who build, test, refine, and share

• LSC’s grantee network is not hesitating in the face of AI, programs across the country are experimenting with tools to reduce administrative burden, streamline intake, and increase organizational capacity

• Legal aid organizations are a proving ground for responsible AI: high-stakes, resource-constrained environments where accuracy, trust, and ethics are non-negotiable

• Developers should build with legal aid, not just for it, engaging practitioners early and often leads to better, more usable, more trusted tools

• AI represents an evolution, not a replacement; it builds on prior innovations like computers, the internet, and e-filing, but moves faster and has lower barriers to entry

• LSC’s eviction report reveals how dramatically local laws shape eviction outcomes, and underscores that eviction, representing 35% of LSC grantee caseloads, has cascading consequences for housing, employment, education, and health

• Landlords are potential partners in housing stability, not just adversaries. Eviction is often a poor business model, and AI can help facilitate alternatives

• LSC secured $540 million in congressional appropriations for FY2026 despite significant political headwinds, reflecting broad bipartisan support for civil legal services

• Collaboration is a multiplier: the legal aid community’s existing culture of sharing and learning from one another is a strategic advantage in adopting AI

Notable Quotes:

1. “The question is no longer whether AI will shape the future of legal services. It already is.” — Ron Flagg (00:01:06–00:01:14)

2. “It’s the innovators rather than the innovation itself that hold the key. It’s a community that builds, tests, refines, and shares for the benefit of all.” — Ron Flagg (00:03:44–00:03:53)

3. “What we’re seeing across the country is not hesitation, but hunger.” — Ron Flagg (00:05:17–00:05:24)

4. “Legal Aid is not just a user of technology. It’s a proven ground.” — Ron Flagg (00:07:12–00:07:20)

5. “In the immediate moment, we need to think about integration, not disruption. How does a tool fit into existing workflows? How does it reduce the burden of those workflows rather than adding to it?” — Ron Flagg (00:15:33–00:15:49)

6. “What other economic endeavor in America has as a model throwing out your customers?” — Ron Flagg (00:26:34–00:26:45)

7. “Collaboration is a multiplier. Our field already does it well. Build on that collaborative nature. Stay focused on purpose. This is not innovation for its own sake — it’s about making our justice system more accessible and more fair.” — Ron Flagg (00:30:18–00:30:35)

8. “If we get this right, we won’t just be building better tools. We’ll be building a stronger, more capable system of justice — one that meets people where they are and delivers on the promise that justice should be accessible to all.” — Ron Flagg (00:09:07–00:09:27)

Clips

Landlords, Don’t Evict Your Customers

Build for Integration, Not Disruption

AI Where It Actually Happens

Why Eviction Laws Matter

Ron’s perspective is shaped by decades at the intersection of law, public service, and institutional leadership, and what comes through most clearly is his refusal to accept the status quo. The justice gap he describes is not abstract: it is half a million people turned away from legal help every year, families losing their homes without representation, veterans unable to access benefits they earned. His call to action is not a lament; it is a blueprint.

What makes this conversation especially timely is Ron’s framing of AI not as a threat to legal services, but as a long-awaited force multiplier for a field that has always done more with less. The legal aid community, he argues, has been innovating under constraint for decades. AI is simply the next tool in that tradition, and if deployed intentionally, it could be the most powerful one yet.

Closing Thoughts

Sitting across from Ron Flagg, even in a conference setting, you feel the weight of what he carries. He leads an organization responsible for ensuring that the most vulnerable people in America have access to the legal system. And he does it with a combination of intellectual rigor, institutional savvy, and genuine humility that I find rare in any leader, let alone one navigating the political pressures he faces.

What stayed with me most from this conversation is his insistence that the innovators matter more than the innovations. We spend so much time in the legal tech world talking about tools, which AI model is best, which workflow to automate, which firm is deploying what. Ron reminded me that the real work is about community. It’s about people choosing to collaborate, to share what works and what doesn’t, to build with clients and not just for them.

For our Legal Rebels community, Ron’s message is a challenge and an invitation. If you’re building technology, are you building it with the people who will actually use it? If you’re practicing law, are you leaning into experimentation even when it’s uncomfortable? If you’re a researcher or funder, are you helping the field measure what actually matters?

The justice gap is real. It is large. And it will not close by accident. But as Ron made clear, we have the community, the culture, and now the tools to make a real dent in it. This is our year to build, and I hope this episode inspires you to build something that lasts.

If you want to understand how AI is reshaping the future of dispute resolution and why lawyers must be part of building that future, you need to listen to this episode. Bridget is at the forefront of AI-native dispute resolution and brings a uniquely visionary yet grounded perspective on what’s possible when mission-driven institutions embrace exponential change.

AI, Access, and the Architecture of What Comes Next

Join me as I share Bridget McCormick’s keynote from the LawDroid AI Conference 2026, delivered live to our community of legal innovators.

In this powerful keynote, Bridget, President and CEO of the American Arbitration Association, lays out a sweeping and inspiring vision for the future of dispute resolution. She shares how the AAA became the first ADR organization in the world to offer AI-native dispute resolution, and what the journey of building that looked like from the inside. From deploying a fully agentic AI arbitrator to building a resolution simulator and an AI-native case management platform, Bridget walks us through how a 100-year-old nonprofit is reimagining what justice delivery can look like.

Her talk is grounded in data, rooted in the AAA’s founding mission, and animated by a genuine belief that AI’s best legal use case may be dispute resolution itself. This is a must-watch for anyone who cares about access to justice, the future of legal institutions, and what it means to lead with imagination rather than compliance in a moment of exponential change.

The Skinny

Bridget McCormick, President and CEO of the American Arbitration Association, delivered the opening keynote at the LawDroid AI Conference 2026, offering a sweeping and data-rich vision for AI’s role in the future of dispute resolution. Drawing on three years of leading the AAA’s AI transformation, Bridget traces the organization’s journey from early internal experimentation to launching the world’s first AI-native dispute resolution product, a fully agentic AI arbitrator currently operating in document-only construction disputes. Anchored in the AAA’s founding mission (democratized access to dispute resolution for individuals) not just institutions, Bridget connects the idealism of founder Frances Kellor to the urgent opportunity AI represents today. She challenges the legal profession to stop waiting for task forces and white papers, and instead take a seat at the table where the future is being built, with or without lawyers.

Key Takeaways:

• The AAA became the first ADR organization in the world to offer AI-native dispute resolution, launching a fully agentic AI arbitrator for document-only construction disputes in November 2025

• Bridget’s four-part framework for navigating exponential change: ground decisions in data, anchor to mission, move fast because action leads to information, and treat AI as a platform shift, not an incremental improvement

• A recent MIT study found 90% of employees had used AI at work while only 40% of their companies had purchased an AI solution, employees aren’t waiting for permission

• According to Law360, the share of lawyers using AI for at least one purpose jumped from 35% in 2024 to 54% in 2025, and frequent users are dramatically more optimistic than non-users

• The AAA’s AI arbitrator uses a fully agentic architecture where agents parse claims and evidence, show their work to the parties, and iterate until both sides feel heard, a breakthrough for procedural fairness

• The AAA is also building a resolution simulator, an AI-assisted mediation system, and a fully AI-native case management platform, expected to be live within two years

• Bridget invokes the AAA’s founder, Frances Kellor, who created the organization after being disappointed by international institutions that failed to prevent World War I — as the North Star for its mission: democratized, individual-level access to dispute resolution

• Three out of four state court cases involve at least one self-represented party, and Bridget sees AI as the most promising catalyst for closing that gap

• Tech innovation follows a pattern of hobbyists → automators → innovators, and Bridget believes we are now entering the innovator phase, where entirely new forms of legal service delivery become possible

• The legal profession risks being “Ubered” out of its own domain if lawyers don’t take a seat at the table where AI is being built

Notable Quotes:

1. “Building in the middle of exponential change can be extremely paralyzing. We move fast because we believe that action leads to information. We’re not waiting for a white paper.” - Bridget McCormick (00:02:15-00:02:58)

2. “The deeper I went, the more I realized this was not another wave of legal tech. This was a platform shift.” - Bridget McCormick (00:06:11-00:06:16)

3. “If there’s one thing that matters in any dispute resolution process, it’s that the parties feel heard.” - Bridget McCormick (00:06:53-00:07:01)

4. “I think AI’s best legal use case may be dispute resolution. At its core, we have two parties asking for a neutral reasoned decision from a process that’s fair, consistent, and affordable. They want a resolution. They don’t want lawyers or arbitrators or judges. We’re just the tools they have right now to get there.” - Bridget McCormick (00:19:15-00:19:41)

5. “The mission is the root of all of our work. We’re lucky that as a mission-based nonprofit, we had a clear North Star.” - Bridget McCormick (00:27:13-00:27:37)

6. “Our civil justice system is failing most people. Three out of four state court cases involve at least one self-represented party — not by choice, and in most cases it’s because people can’t afford it. I can’t think of a bigger threat to the rule of law than that.” - Bridget McCormick (00:31:14-00:31:37)

7. “The technologists can Uber right through our regulatory barriers if we cede the territory by failing to take a seat at the table.” - Bridget McCormick (00:36:43-00:36:51)

8. “That future is not promised. It’s ours as long as we’re willing to lead, as long as we’re willing to be in the conversation, as long as we’re willing to be at the table.” - Bridget McCormick (00:36:09-00:36:20)

Clips

Inside the AI Arbitrator

AI Isn’t an Incremental Shift

AI Adoption Exploded in Enterprise

AI Could Reduce Conflict

Bridget’s keynote stands out for the clarity with which it connects past to future. The AAA’s founding story, Frances Kellor building a democratized dispute resolution institution after the failures of international arbitration to prevent World War I, becomes more than historical context. It becomes the lens through which every AI investment the AAA makes is evaluated. That kind of mission alignment is rare, and it’s precisely what allows an organization to move fast without losing its way.

What makes Bridget’s vision especially compelling is her insistence that continuous improvement is not enough. Automating what we already do better is valuable — but it is the innovator phase, where entirely new forms of justice delivery become imaginable, that holds the real promise. The AI arbitrator isn’t just a faster version of what arbitrators do. It’s a fundamentally new architecture for helping people feel heard, understood, and fairly resolved.

Closing Thoughts

Bridget McCormick’s keynote was one of the highlights of the LawDroid AI Conference 2026, and listening to it again, I’m struck by how much she managed to pack into a single talk: data, history, product vision, and a genuine moral urgency that you don’t often hear from the stage at legal tech conferences.

What resonates most with me is her point about the innovator phase. For years, the conversation in legal tech has been dominated by automation, doing the same things faster and cheaper. That’s important. But Bridget is pointing at something bigger: the moment when entirely new forms of legal service delivery become possible. The AI arbitrator isn’t just a more efficient version of arbitration. It’s a new art form, to use her analogy, like cinema emerging from the innovation of the movie camera.

For our Legal Rebels community, the message here is both inspiring and urgent. The future of dispute resolution, and legal services more broadly, is being built right now, largely by people who are not lawyers. Bridget’s call to action is direct: we need to be at the table. We need to be part of the architecture of what comes next. Because if we’re not, the technologists will build it without us, and it may not reflect the values of fairness, access, and procedural justice that our profession, at its best, is supposed to stand for.

Frances Kellor believed that giving individuals the tools to resolve their own disputes could change communities, countries, and the world. A hundred years later, we finally have technology powerful enough to make that vision real. That’s the moment we’re in, and I don’t intend to miss it!

Episode Summary

In this month’s AI Double Take, LawDroid CEO Tom Martin and Chief Legal Futurist Sateesh Nori record live off the energy of the LawDroid AI Conference 2026 — a two-day virtual event that brought together the legal tech community under the banner of “The Year to Build.” They unpack standout moments from three keynotes, including Nikki Shaver’s bullish legal tech market analysis and Bridget McCormick’s call to stop incrementalising and start transforming. The hosts also dig into a striking MIT study showing AI-driven court filings are up 5%, the chronic underfunding of legal services relative to military spending, and what it would take for Big Tech to meaningfully close the access to justice gap. The throughline: the legal system is at an inflection point, and the only wrong move is to stand still.

Key Takeaways

1. LawDroid AI Conference 2026 — A Community Comes Together

The third annual LawDroid AI Conference ran April 28–29 as a fully virtual event with three keynotes, workshops, and panels covering the real-time pulse of AI in law. Sateesh served as MC across both days. The hosts describe the experience as reinvigorating — a reminder that a serious, optimistic community is already building the next chapter of legal services.

2. Nikki Shaver’s Keynote — No Bubble, Just a New Ceiling

Nikki Shaver, co-founder and CEO of Legal Technology Hub, presented data showing an upward trajectory in legal tech that shows no sign of reverting. Rather than a dot-com-style bubble, the data points toward a breakout: AI enabling legal services to finally reach the vast latent legal market — the millions of people currently unserved. The image Tom offered: Willy Wonka’s glass elevator breaking through the ceiling and kept going.

3. The Velocity Warning — When Outside Outpaces Inside

Tom introduced the counterpoint: when the velocity of innovation outside an organisation exceeds the velocity of change inside it, that organisation is in trouble — the Sears, Blockbuster pattern. Sateesh framed it as the classic innovator’s dilemma: established legal players have no structural incentive to disrupt themselves. The risk is that hungrier, smaller, tech-native players eat the market while incumbents wait.

4. Bridget McCormick’s Keynote — Stop Improving the Candle

Bridget McCormick, President of the American Arbitration Association, delivered a pointed message: enough with committees, commissions, and incremental improvement. Her centrepiece: you cannot continuously improve a candle into a light bulb. It is a categorically different thing. The legal field needs to stop asking for faster horses and start imagining the automobile. Her call to action aligned directly with the conference theme: build now.

5. Ron Flagg / LSC Fireside Chat — Funded But Under-Resourced

In a fireside chat with Sateesh, Legal Services Corporation President Ron Flagg reported that LSC has maintained current funding levels despite severe political headwinds — a genuine achievement. But the scale of the gap remains stark: the entire annual LSC budget ($566 million) amounts to less than a quarter of a single day’s military spending. Both Ron Flagg and former LSC President Jim Sandman are publicly calling for legal aid organisations to accelerate AI adoption. The signal from the top is clear.

6. AI Could Automate Over 70% of Access to Justice Work

A post by Richard Truman of Artificial Lawyer, using AI-based modelling, estimated that over 70% of access to justice work is amenable to automation — significantly higher than the 30%+ figure for large law firms. For legal aid organisations operating without the billable hour incentive, the financial case for AI adoption is structurally aligned in a way it simply is not for Big Law.

7. The Point One Challenge — Big Tech’s Untapped Obligation

Sateesh proposed the “Point One Challenge”: if Big Tech companies contributed just 0.1% of annual profits to access to justice, it would generate roughly $5 billion per year. Tom raised the complementary idea of noblesse oblige — voluntary goodwill investment from Anthropic, Microsoft, and others who are already entering the legal market. Neither idea is mainstream yet, but both would dwarf current public funding.

8. The Direct-to-Consumer Alternative — Agency Over Charity

Sateesh argued that waiting on Big Tech goodwill misses a more powerful model: direct-to-consumer legal tools that give individuals agency. People do not want charity — they want to pay a dollar for a legal answer or five dollars for a form. LawDroid is actively developing in this direction, and Tom confirmed something may be visible in the near future.

9. AI Filings Up 5% — Courts Must Adapt, Not Resist

MIT researchers Shaw and Levy studied federal civil case filings and found a 5% increase in the post-AI period, with significant evidence of AI-generated content in the filings themselves. The hosts’ joint position: this is not a crisis to suppress — it is a signal that people with legitimate claims are finally able to access the system. Courts should lean in with their own AI tools (such as the Learned Hand system already in use by some judges), offer front-end guidance on case viability, and scale to handle volume rather than shutting the door on access.

Show Notes

Topics Covered

• LawDroid AI Conference 2026 recap (April 28–29, virtual, “The Year to Build”)

• Nikki Shaver keynote: legal tech market trajectory and the latent legal market

• Dot-com bubble concerns vs. breakout growth scenario

• The velocity of outside innovation vs. inside change (Welch/Darth Vaughn quote)

• Sears and Blockbuster as cautionary tales for legal incumbents

• Bridget McCormick keynote: candle-to-light-bulb transformation imperative

• Henry Ford “faster horses” analogy applied to legal transformation

• Ron Flagg / LSC fireside chat: current funding levels, political headwinds

• LSC budget ($566M) vs. daily military spending comparison

• Jim Sandman’s public call for AI adoption across LSC organisations

• Richard Truman / Artificial Lawyer: 70%+ of access to justice work automatable

• Big Tech entering legal AI: Anthropic, Microsoft, Elon Musk / Grok

• Anthropic’s 20,000-person Claude legal webinar (surprised by demand)

• Noblesse oblige / Big Tech goodwill funding for access to justice

• Sateesh’s “Point One Challenge” — 0.1% of Big Tech profits = ~$5B/year

• Direct-to-consumer legal tools: agency over charity model

• MIT study (Shaw and Levy): federal civil filings up 5% post-AI

• AI-generated filing detection and court adaptation strategies

• Learned Hand AI system used by judges

• LawDroid upcoming product hint

People & Organizations Mentioned

• Tom Martin — CEO & Founder, LawDroid

• Sateesh Nori — Chief Legal Futurist, LawDroid; MC at LawDroid AI Conference 2026

• Nikki Shaver — Co-founder & CEO, Legal Technology Hub; Day 2 keynote

• Bridget McCormick — President, American Arbitration Association; Day 1 keynote

• Ron Flagg — President, Legal Services Corporation; fireside chat guest

• Jim Sandman — Former President, Legal Services Corporation

• Richard Tromans — Artificial Lawyer (legal tech blog/newsletter)

• Hannes Westermann — Legal AI researcher (referenced re: Anthropic legal webinar data)

• Shaw & Levy — MIT researchers; federal civil filing AI study

• Anthropic — Ran a 20,000-attendee Claude legal webinar; entering legal market

• Microsoft — Noted as entering legal AI market

• Elon Musk / Grok — Posted that Grok is “#1 for AI law”

• Legal Services Corporation (LSC) — Primary federal funder of civil legal aid in the US

• American Arbitration Association — Led by Bridget McCormick; dispute resolution

• Legal Technology Hub — Legal tech market intelligence, led by Nikki Shaver

• Learned Hand — AI tool used by judges for managing case volume

Final Takes

Sateesh Nori:

“I feel really optimistic after our conference. Seeing all the folks we might run into here and there — but together in one place, in our forum — that was really great. It gives me a lot of hope. There are a lot of good people in this space. Really funny, really interesting people. I’m really glad to be part of this community.”

Tom Martin:

“It is so reinvigorating every year to have the conference, to have people attending, to feel that sense of community come together. I feel energised for this year. We’re going to accomplish a lot together, and I’m really excited about that. We’re not always familiar with the destination we’re going — but it looks like it could be a good place for more of us. So let’s all keep building.”

AI Double Take is produced by LawDroid | lawdroid.com

If you missed the LawDroid AI Conference 2026 last week, you need to listen to episode! It’s Nikki Shaver’s keynote about what is actually happening in legal AI right now; not the marketing version, but the real one from someone who knows. Nikki has the clearest view of the legal technology landscape of anyone in our field, and she brought that perspective in full force. Don’t miss it!

The State of Legal AI 2026: What’s Now, What’s Coming

Join me as I share Nikki Shaver’s landmark keynote from the LawDroid AI Conference 2026.

In this powerful keynote, Nikki Shaver, co-founder and CEO of LegalTech Hub and one of the sharpest minds in global legal technology intelligence, delivers an unflinching look at where the legal AI market truly stands. Drawing on years of tracking every AI solution in the legal market, Nikki walks us through the explosion of legal tech startups, the dramatic rise of agentic AI, the surge of private equity and Silicon Valley capital into legal, and the structural shifts now reshaping law firms at their core.

Her keynote doesn’t just catalog what’s happening. It asks the harder question: given all of this change, what must law firms actually do to survive and thrive? Nikki introduces her “New Horizons” framework for differentiation, explains why AI enablement is now just table stakes, and warns that the firms not yet thinking beyond Horizon One are already behind. This is essential listening for any legal professional who wants to understand the moment we are in, and what to do about it.

The Skinny

Nikki Shaver, founder of LegalTech Hub, delivered the opening keynote of the LawDroid AI Conference 2026 with a sweeping, data-rich assessment of where legal AI stands today. Speaking from a unique vantage point, advising law firms, corporate legal departments, vendors, and investors on a weekly basis across multiple continents, Nikki pulled back the curtain on the real signal beneath all the noise. The legal tech market has seen approximately 100 net new startups per quarter since 2023 with no signs of slowing, legal tech funding is outpacing all prior years, and agentic AI has moved from curiosity to concrete use cases in legal workflows. Perhaps most striking is Nikki’s assessment that the legal industry has entered the “Innovator’s Dilemma zone,” a moment where incumbent law firms are still thriving on their existing models even as nimble, AI-native disruptors are scaling rapidly around them. Her central message: AI enablement is now table stakes, not a differentiator. True survival requires firms to think in terms of her New Horizons model, moving beyond efficiency gains toward transformed systems, embedded legal infrastructure, and scalable outcome-based offerings.

Key Takeaways:

• LegalTech Hub has tracked approximately 100 net new legal AI startups per quarter since early 2023, with the market showing no signs of a bubble bursting

• Agentic AI in legal has exploded from roughly 100 solutions in July 2025 to a dramatically larger ecosystem by March 2026, signaling a shift from task-based to workflow-based AI

• Between 79 and 90% of lawyers are now using some form of AI in practice, with corporate legal department adoption swinging from 23% to 52% in a single year

• Legal professionals anticipate saving approximately 240 hours annually using AI, representing a potential $32 billion annual impact on the US legal sector alone if work is not repriced

• Legal tech has gone fully mainstream: celebrity endorsements, sports sponsorships, and mass-market advertising are now part of the landscape

• Anthropic’s entry into legal with Claude Legal Skills created significant market anxiety, though Nikki’s conversations with Anthropic suggest it was intended as a model marketing play, not a direct product launch

• Law firm mergers jumped 21% in the first half of 2025, and private equity investment in law firms is growing via new managed services structures that navigate non-lawyer ownership rules

• AI-native law firms, built on the operational foundation of AI from day one, represent a new category of disruptor unlike anything the industry has seen before

• Nikki’s “New Horizons” model reframes the McKinsey Three Horizons framework for the AI era: Horizon One is efficiency (already table stakes), Horizon Two is quality and augmented judgment, and Horizon Three is fully transformed, scalable, embedded legal systems

• Firms focused solely on Horizon One innovation right now are already at risk of falling behind in what Nikki calls the “dilemma zone”

• True differentiation requires leveraging your firm’s unique data, building governance capabilities, investing in R&D, and structuring for work that doesn’t look like traditional matters

Notable Quotes:

1. “If you want to know what’s actually happening in legal AI in 2026, not the marketing version, the real one, Nikki is the person to start the day with.” - Tom Martin (00:00:32-00:00:40)

2. “On a weekly basis, I am speaking with law firms, often in different parts of the world. I speak to corporate legal departments. I speak to vendors, both very early stage and very mature in the market. And I speak to investors as well as law students. I teach at a law school. And so I get a very well-rounded view on what is happening in the legal market.” - Nikki Shaver (00:02:03-00:02:31)

3. “We have not seen that. In fact, we’ve seen the opposite. There’s just an enormous proliferation of startups still over and above consolidation in the market, even though we’ve seen more M&A in the market as well over the past year.” - Nikki Shaver (00:06:22-00:06:35)

4. “Already by now April in 2026, I can say with certainty that we’re finding a lot of use cases for agentic AI. And generally speaking, the dialogue around AI and legal has shifted to agent.” - Nikki Shaver (00:09:04-00:09:18)

5. “When the rate of change on the outside exceeds the rate of change on the inside, the end is near.” - Jack Welch, quoted by Nikki Shaver (00:22:50-00:22:58)

6. “It is now table stakes to have AI enablement across the firm. The technology that all of us rush to license is just the cost of entry at the moment. It’s not where you differentiate.” - Nikki Shaver (00:39:20-00:39:35)

7. “We do not have that time anymore.” - Nikki Shaver, on the traditional Three Horizons timeline (00:40:53-00:40:56)

8. “If it’s not now, I don’t know when it will be.” - Nikki Shaver (00:49:13-00:49:15)

Clips

Anthropic Shocked the Legal Market

Law Firms: Your Structure is Killing Innovation

AI is Now Table Stakes

PE, Regulation, and Clients on AI

Nikki’s keynote is remarkable not just for the data it presents, but for the clarity of the narrative it constructs around that data. The legal tech market is not in a bubble. It is not slowing down. The pace of startup activity, the volume of funding, the acceleration of adoption, and the entry of new classes of competitors, from AI-native law firms to tech-product companies pivoting into legal services, all point in one direction. The window for comfortable, incremental change has closed.

What makes Nikki’s perspective especially valuable is that she doesn’t traffic in fear for its own sake. Her New Horizons model is fundamentally optimistic: there is a path forward, but it requires law firms to be honest about where they actually are, to invest in R&D the way every other industry does, to leverage their own data as a strategic asset, and to hire for the kind of creative, outside-the-box thinking that the current moment demands.

Closing Thoughts

As someone who has been in the legal technology space for years, building tools, talking to lawyers, and watching this industry evolve, I found Nikki’s keynote to be one of the most grounding and clarifying things I’ve heard in a long time. Not because it was comfortable, but because it was honest.

We are in the dilemma zone. I’ve felt it. Many of you have felt it. The pace of change is outrunning our ability to process it, let alone respond to it. And Nikki’s data makes clear that this is not a temporary condition that will settle down once the hype fades. The hype isn’t fading. The startups keep coming. The capital keeps flowing. The clients are now asking, sometimes demanding,?that their law firms use AI.

What struck me most was the point about table stakes. So many firms have spent the last two years treating AI adoption as a differentiator, as a competitive edge. Nikki’s data says that moment is over. Everyone has Harvey or Legora or something similar. The question now is what you do with your own data, your own expertise, your own client relationships, how you build something that can’t be replicated by the next firm that licenses the same tool.

For our Legal Rebels community, that is both the challenge and the opportunity. The firms and professionals who will thrive are those who don’t wait for the market to settle, who invest in building real capabilities now, and who are willing to reimagine what legal services can look like when AI is at the core of everything, not bolted on at the edges.

Now is the time. As Nikki said: if not now, when?

Have you ever felt out of sync with the AI you work with? That it does not quite get you the way another model used to? Maybe you think fondly of the easy back-and-forth you used to have with GPT-4o, before it was quietly put out to pasture, and the model that replaced it, for all its raw capability, just is not the same. The drafts come back a half-step off. The prompts that used to land now glance off. The rhythm has gone, and you cannot quite say why.

If that is your experience, you are not alone, and the explanation is probably not the one you have been reaching for. We tend to blame the model, or blame ourselves, or blame the prompt. “It’s not you, it’s me.” Most of the time, however, none of those is the real culprit.

The real culprit is fit.

The real question is whether the you and your AI are a good fit.

I call this The Jagged Fit.

AI capability has an irregular contour, brilliant at some tasks, hopeless at adjacent ones. Human ingenuity also has an irregular contour, brilliant at some things, hopeless at adjacent ones. The two surfaces can lock together or they can grind against each other. The pairing, not the person and not the model, is the measure of capability.

If this sounds interesting to you, read on...

This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.

The Jagged AI Frontier

Ethan Mollick gave us the metaphor that started this. He called it the jagged frontier of AI capability. Picture a coastline, irregular and unpredictable. On one side of the line, the model is genuinely capable. On the other side, it fails, sometimes catastrophically. The line itself does not run where you expect it to run. Tasks that look hard turn out to be easy for the model. Tasks that look easy turn out to be hard. You cannot tell from the outside which is which.

The current generation of frontier models is remarkable. They can extract information and produce structure from unstructured text at speeds no human can match. They draft, summarize, translate, classify, and pattern-match across enormous bodies of material. For a profession that has been buried in documents for centuries, this is not a small thing.

But the jagged frontier is real. Inside the same conversation, a model that has just produced a useful contract analysis will confidently invent a citation. A model that has summarized a 90-page deposition with insight will fail at a piece of basic arithmetic. The capabilities of AI are like a coastline, with bays and inlets and the occasional cliff. Mollick’s contribution was to give us a way to see this clearly. AI is not uniformly competent or uniformly incompetent. It is jagged.

The natural conclusion most people draw from this is: learn the coastline. Find out where the model is strong and where it is weak, and stay on the strong side of it. That is good advice as far as it goes. But it stops one move short of the more interesting question and acknowledgment.

Humans Are Jagged Too

Psychology has been telling us this for a century. The single-number version is IQ. One score, one point on a bell curve, one ranking. IQ has been useful for some things and disastrous for others. It is contaminated by cultural assumptions baked into the tests. But, the deeper issue with IQ is it pretends intelligence is one dimensional.

Howard Gardner’s theory of multiple intelligences, whatever its empirical limits, points toward different dimensions: linguistic intelligence, logical-mathematical intelligence, spatial intelligence, musical intelligence, interpersonal intelligence, intrapersonal intelligence, kinesthetic intelligence. People are not equally strong across these dimensions. They are not even comparable across them. A great trial lawyer and a great patent lawyer are drawing on different intelligences, and the patent lawyer might be lost in front of a jury while the trial lawyer might be lost in describing an invention. Both are excellent. But, their excellence has a different shape.

Personality has the same structure. The Big Five model gives us five independent axes: openness, conscientiousness, extraversion, agreeableness, neuroticism. Two people can have the same overall “score” and be unrecognizably different in how they work, what they notice, what frustrates them, what energizes them. Cognitive style is similar. Some people think in pictures, some in words, some in systems. Some need silence to concentrate. Some need pressure. Some are at their best at 6 a.m. Some are night owls.

Take the Myers-Briggs types as accessible shorthand, mindful of the instrument’s well-known limits. Picture an INFJ lawyer and an ISTJ lawyer working the same case. The INFJ thinks in patterns and meanings. She sees the matter as a story before she sees the facts as a list, and she is uneasy until the narrative comes together. The ISTJ thinks in ordered facts. He sees the matter as a record before he sees it as a story, and he is uneasy until the timeline is airtight and the citations are all squared away. Both can be excellent lawyers. Both can sit at desks across the hall from each other. Their best work has different shapes, and the work that drains one is often the work that energizes the other. Hand them the same memo to draft and you will get two unrecognizable products, each defensible, each shaped by the mind that produced it.

When you put all of this together, the picture is unavoidable. Human intelligence, like AI capability, is jagged. Each of us has a coastline. Each of us has bays where the work flows and cliffs where it stalls. The jaggedness is not a flaw to be smoothed; it is a feature of being a unique individual.

When Two Jagged Edges Meet

Now place the two maps side by side.

If you press two jagged edges together at random, they clash. There are gaps where neither side fills the space, and ridges where both sides claim the same territory. This is the bad fit. The lawyer’s strength overlaps with the model’s strength, so neither is leveraged. The lawyer’s weakness overlaps with the model’s weakness, so neither is covered. The pairing produces less than either party would produce alone.

But if you align two irregular surfaces with attention to their contours, something different happens. The peaks of one fit the valleys of the other. The pieces interlock. The lawyer’s weakness is met by the model’s strength; the model’s weakness is met by the lawyer’s strength. The pair becomes more capable than either party alone, and the increase is not modest. It is the difference between dragging a tool and dancing with a partner.

Consider two associates. The first associate has a strong intuitive sense for narrative and a weak appetite for procedural detail. Pair her with a model that is rigorous about citations, exhaustive in checking deadlines, and willing to fact-check her drafts without complaint. Her weak side is covered. Her strong side is amplified, because she now has time to do what she is best at. The fit works.

The second associate is the inverse. He is meticulous with detail, careful with the record, and slow to commit to a narrative arc. Give him the same model and you have doubled down on his strength and left his weakness exposed. He needs a different partner: one tuned to push him toward narrative commitment, to surface the human story buried in the documents, to draft boldly so that he can edit. Same firm, same case type, different fit.

This is why “which AI is best for lawyers” is the wrong question. The right question is which AI is best for this lawyer, on this kind of work, in this phase of her career.

I want to be careful because the dating metaphor in the title can be taken too far. I am not arguing that AI tools have personalities in any deep sense, or that we should anthropomorphize them. I am saying something different. Different models, with different training, different defaults, different reasoning styles, behave differently. Different lawyers, with different intelligences, different temperaments, different working styles, work differently. When you pair an AI with a lawyer, the combination has capabilities neither has on its own. Some work. Some don’t.

The Matchmaking Problem

Organizational psychology has worked on a version of this problem for fifty years under the name person-environment fit. The literature is large and the findings are robust: when a person’s strengths, values, and working style align with the demands and culture of their role, performance and well-being both rise. When they misalign, performance drops and burnout follows.

The same logic applies to person-AI fit. On the human side: cognitive style, domain expertise, personality profile, the specific tasks the lawyer actually performs in a typical week, the kinds of mistakes she is prone to making, the kinds of work that energize her, the kinds that drain her. On the AI side: model behavior under different prompt styles, default tone, willingness to push back, depth of reasoning, hallucination patterns by domain, latency, the shape of its strengths and weaknesses across the practice areas in question. None of this is mysterious. Most of it is measurable. We have not measured it yet in any organized way because we are still treating AI procurement as a software decision rather than as a partnership decision.

In machine learning, every serious model now ships with what its developers call a model card: a document describing the model’s intended uses, training data, performance characteristics, and known limitations. The cards exist precisely because models are not interchangeable. Read three cards side by side and you start to see why fit matters.

Model A, the Cautious Generalist. Strong at nuanced writing, careful with citations, willing to flag its own uncertainty, capable of sustained reasoning over long documents. Slower to commit to a thesis. Hedges when pressed. Best fit: lawyers who already have strong views and want a partner that will test them.

Model B, the Confident Synthesizer. Fast first drafts, broad general knowledge, willing to commit to a structure or a position. Confident even when wrong. More prone to inventing citations under pressure. Best fit: lawyers with deep domain expertise and strong editorial instincts who need a fast starting draft to react against.

Model C, the Citation-Anchored Specialist. Grounded in retrieval, will not invent cases, careful with procedural detail, deep in regulated practice areas. Narrow outside its domain. Less fluent at narrative or argument. Best fit: lawyers in heavily regulated areas where errors are catastrophic and retrieval discipline is the central virtue.

Now hand each of those models to the INFJ and the ISTJ from earlier and watch what happens. The INFJ paired with Model C is in a good marriage. The ISTJ paired with Model B is in a dangerous one. The INFJ paired with Model A may have found a sparring partner that finally pushes her ideas into shape. The ISTJ paired with Model A may find his work bogged down. The model cards and the personality types are crude tools, and yet even with crude tools the matching question becomes clear.

I do not think we are far from being able to do this matching with AI, especially as models become more powerful. The signals are there. The instrumentation is there. What is missing is the recognition that the question matters.

Closing Thoughts

The first generation of AI in law has been dominated by a question of the form: which model is best? I think it’s the wrong question, the way “what is the best food” is the wrong question. It assumes a universal palate. There is no universal taste. There is no universal lawyer either.

The second generation will be dominated by a different question. Not which model, but which pairing. Not capability, but fit. The lawyer who flourishes with AI will not necessarily be the most technical or the most enthusiastic. She will be the one who has found, by luck or by design, an AI partner whose jagged edges meet hers.

The technology will keep getting better. The frontier will keep moving. The coastline will keep shifting. None of that changes the underlying point, which is that two irregular shapes, well matched, can do what neither could do alone.

So if your AI is not that into you, it might not be your fault. It might be that you are simply with the wrong partner.

The right fit is out there. Play the field.

Tom Martin is CEO & Founder of LawDroid, Adjunct Professor at Suffolk University Law School, and Author of AI with Purpose: A Strategic Blueprint for Legal Transformation (Globe Law and Business). He is “The AI Law Professor” and writes his eponymous column for the Thomson Reuters Institute.

If you want to understand how empathy, storytelling, and technology can converge to make legal services accessible to people who need them most, you need to listen to this episode. Jonathan is at the forefront of access-to-justice innovation and brings a decade of hard-won, on-the-ground experience to this conversation.

Turning Empathy Into Access: How Upsolve Is Rewriting the Rules of Bankruptcy Law

Join me as I interview Jonathan Petts, co-founder and CEO of Upsolve.

In this insightful podcast episode, Jonathan shares his journey from the Upper East Side of Manhattan to the clerk’s offices of Brooklyn bankruptcy court , and ultimately to building one of the most impactful legal tech nonprofits in the country. He dives deep into how Upsolve has evolved from a manual spreadsheet operation into an AI-powered platform that helped 115% more people get a fresh financial start last year alone. Jonathan also walks us through how their AI Paralegal tool is deflecting 50% of support questions and enabling each paralegal to serve four times the number of people they once could.

His stories and insights reveal what it truly takes to close the access-to-justice gap, not with theory, but with practical, iterative, deeply human-centered design. This episode is a must-watch for anyone who believes that technology should serve people first, offering a powerful model for what legal innovation looks like when empathy leads the way.

The Skinny

Jonathan Petts, co-founder and CEO of Upsolve, has spent the last decade building what he describes as a “TurboTax for bankruptcy” for low-income Americans who can’t afford lawyers. In this conversation, Jonathan traces his path from growing up on Manhattan’s Upper East Side, shaped by an English father who prized education and a California mother who modeled social ease, through an English degree at the University of Pennsylvania, to law school and a pivotal summer interning in a Brooklyn bankruptcy court clerk’s office during the chaotic lead-up to the 2005 bankruptcy law reform. That experience, watching overwhelmed people wade through 100 pages of dense legalese with Staples-bought form packets, left him both inspired and angry at how unnecessarily hard the system made it to access a constitutional right. After being let go from two large law firms, Jonathan found his footing by returning to serve, eventually co-founding Upsolve with Harvard undergrad Rohan Pavluri after a chance meeting at a pro bono lunch. A Y Combinator alum, Upsolve has now helped thousands find a fresh financial start, and with the addition of an AI Paralegal tool, the organization saw the number of people each paralegal could serve quadruple last year, with users of the AI finishing the process at three times the rate of those who don’t use it.

Key Takeaways:

• Upsolve’s AI Paralegal tool now deflects approximately 50% of questions that previously required human paralegal response, allowing staff to serve dramatically more people

• Users who engage with the AI Paralegal complete the bankruptcy filing process at a three times higher rate than those who don’t

• Last year, U.S. bankruptcy filings rose 15%, but the number of people Upsolve helped get a fresh start rose 115%, a gap largely attributed to the AI tool and operational improvements

• Jonathan’s formative experience in a Brooklyn bankruptcy court clerk’s office during the 2005 reform period revealed the human cost of bureaucratic complexity and set him on his mission

• The AI Paralegal is context-aware, it understands where a user is in the filing process and tailors its responses accordingly, making support feel less like a chatbot and more like a knowledgeable guide

• Jonathan’s advice to new legal tech builders: think carefully about distribution from day one, because building a solution is now easier than ever, but getting it into the hands of people who need it remains the hardest challenge

• Focus beats expansion: when Upsolve tried to expand to immigration and debt lawsuits simultaneously, their core bankruptcy service suffered, a reminder that doing one thing well is harder than it looks

• Jonathan is candid about his own struggles with work-life balance as a founder with two young daughters, and is planning a three-month sabbatical to mark Upsolve’s 10-year anniversary this summer

Notable Quotes:

1. “We have 56,000 people each year who start this tool and don’t get to the end of it. Part of the reason is because they can’t get the support they need.” - Jonathan Petts (00:01:44-00:01:58)

2. “People that use the AI are finishing at a three times higher rate. All of that gives us a lot of conviction that this AI tool to empower folks to unblock themselves is a really high leverage way to increase access.” - Jonathan Petts (00:06:32-00:06:49)

3. “Why is it so hard? Why do you have to fill out 100 pages of dense legalese in order to access this legal process that’s guaranteed by the Constitution?” - Jonathan Petts (00:18:39-00:18:47)

4. “These forms were designed by well-intentioned experts, but those experts weren’t talking to the folks that I was talking to in Brooklyn who couldn’t afford lawyers. And they weren’t designed with those folks in mind. And that, I believe, is the cardinal sin of so much of our consumer legal system in the U.S.” - Jonathan Petts (00:18:51-00:19:23)

5. “When you have no confidence, the thing to do is to find someone you can serve because service connects you with everything you have to give. And that builds confidence.” - Jonathan Petts (00:21:24-00:21:37)

6. “It’s never been easier to create useful solutions and never been cheaper. And the question is, are you going to be able to get that solution in the hands of the people that need it?” - Jonathan Petts (00:28:33-00:28:45)

7. “Building one good access to justice tool — if it’s not getting better, it’s getting worse. That’s a big learning for me.” - Jonathan Petts (00:32:55-00:33:05)

8. “Every day or every other day, I’ll wake up to a new Google review from someone who I’ve never met, in who knows what part of the country, that talks about how Upsolve guided them through one of the darkest times of their life to a better future.” - Jonathan Petts (00:35:18-00:35:32)

Clips

The Cardinal Sin of Legal Design

Built a Non-Profit TurboTax

AI Tripled Completion Rates

YC Forced Us to Think Big

Jonathan’s story is ultimately a story about what happens when you design legal tools for the people who actually need them. From spending a summer watching desperate filers wrestle with $29.99 Staples packets to building an AI-powered platform that quadruples paralegal capacity, he has never lost sight of the human being on the other end of the process. His approach, start by doing things that don’t scale, learn deeply, then build, is as applicable to legal innovation as it is to any startup.

What stands out most is Jonathan’s insistence on focus. The temptation to expand Upsolve’s model to other legal domains was real, and the pilot revealed a hard truth: doing two things at once can make both worse. That discipline, staying close to the mission, improving incrementally, and letting the impact compound, is what has allowed Upsolve to outpace the rising tide of financial distress in this country and actually move the needle for real people.

Closing Thoughts

Jonathan’s journey is one that I find genuinely moving. He didn’t start from a place of certainty, he was fired from two law firms, told he shouldn’t be a lawyer, and had to rebuild his sense of purpose from scratch. What he found when he returned to serving people was something we talk about a lot in the Legal Rebels community: that the legal profession, at its best, is a calling to help people through their hardest moments.

What strikes me about Upsolve’s AI Paralegal isn’t just the impressive numbers: 50% question deflection, a 3x completion rate, paralegals serving four times as many people. It’s what those numbers represent: real families who got a fresh start because someone decided that the bureaucratic complexity of our legal system wasn’t acceptable, and then built something to fix it.

For those of us building in legal tech, Jonathan’s story is both a challenge and an invitation. The tools have never been more powerful. The access-to-justice gap has never been more visible. And the model Upsolve has pioneered, empathetic design, iterative improvement, ruthless focus, is one any of us can apply in our own corner of the profession.

The future of legal services belongs to those who keep the human being at the center of everything they build. Jonathan Petts and the Upsolve team are showing us exactly what that looks like.

LAST CHANCE! — the LawDroid AI Conference 2026 starts tomorrow: April 28–29, virtual, and completely free — two days of keynotes, panels, and workshops on AI and the legal profession. I’d love to see you there.

Register Now!

Next week, on April 28 and 29, I’m hosting the LawDroid AI Conference 2026, and I want you there!

This is the third year we’ve done this, and every year I’m amazed by the community that shows up. Lawyers, legal aid advocates, court innovators, technologists, professors, and people who simply refuse to be left behind by the changes happening all around us.

Share

This year’s theme is The Year to Build, and the agenda reflects that. We have three keynotes from Bridget McCormack, Ron Flagg, and Nikki Shaver. We have panels on ethics, journalism, access to justice, courts, education, and the practice of law. And on Day 2, we have a full slate of hands-on workshops covering skills engineering, vibe coding, document automation, context engineering, and AI agents.

It’s free. It’s virtual. It’s two days you can attend from anywhere in the world, picking the sessions that matter most to you.

Over 1,000 people have already registered. If you’ve been reading this newsletter, you already know why this matters. Come join the conversation.

Register now at LawDroidAIConference.com

Register Now

See you next week!

Tom Martin

CEO & Founder & Host, LawDroid AI Conference 2026

P.S. - It will be recorded, but you need to register to get access to the recordings.

LawDroid Manifesto is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

If you want to understand how experiential learning can transform the way lawyers think about AI, and why automation bias is one of the most urgent issues facing today’s legal profession, you need to listen to this episode. David is at the forefront of legal technology education and brings a uniquely hands-on, human-centered perspective to how we prepare lawyers for an AI-driven world.

Shaping the Next Generation of AI-Aware Lawyers Through Experiential Learning

Join me as I interview David Colarusso, Practitioner in Residence and co-director of the Legal Innovation and Technology Lab (LIT Lab) at Suffolk University Law School.

In this insightful podcast episode, David shares his remarkable journey from growing up near White Sands in New Mexico, to designing a custom interdisciplinary major at Cornell, to teaching physics with six million YouTube views, to becoming a public defender, data scientist, and now one of legal education’s most inventive minds. He dives deep into how he uses experiential learning, including a brilliantly designed classroom simulation, to teach law students about automation bias and the hidden dangers of over-relying on AI tools.

His stories and insights underscore the critical importance of helping lawyers understand not just how to use AI, but when to question it. This episode is a must-watch for anyone curious about legal education, access to justice, and the human skills that will define the lawyers of tomorrow.

The Skinny

David Colarusso, Practitioner in Residence and co-director of the LIT Lab at Suffolk University Law School, brings one of the most unconventional and intellectually rich backgrounds in legal tech education. A self-described perpetual learner who went from “nature’s laws to people’s laws,” David designed his own major at Cornell combining physics and film, earned a master’s from Harvard’s Graduate School of Education, built a following of over six million YouTube viewers with physics explainer videos, and became a Fulbright exchange teacher in Scotland before ultimately going to law school and serving as a public defender in Massachusetts. It was during his time at the Public Defenders Office, where a massive drug lab scandal forced his entire office to manually search paper files for a chemist’s signature, that David recognized how technology could transform the way the legal system handles data. That insight led him to become the office’s data scientist, then to an adjunct role at Suffolk, and eventually to his current position leading the LIT Lab. In this episode, David walks us through a vivid classroom experiment in which he tricked law students into experiencing automation bias firsthand, revealing how quickly humans learn to over-trust AI tools, and how dangerous that trust can become when those tools fail in ways users don’t anticipate.

Key Takeaways:

• Automation bias, the tendency to over-rely on AI decision tools, is a real and measurable phenomenon that affects even attentive, well-intentioned users, as David’s classroom simulation powerfully demonstrated

• David designed a doc review simulation where students worked with an AI tool that was 100% accurate in early rounds, causing them to stop double-checking, and then performed significantly worse when the tool’s accuracy dropped in later rounds

• Experiential learning is David’s pedagogical superpower: rather than lecturing students about AI risks, he creates simulations that allow students to discover those risks themselves

• David’s path from physics teacher to public defender to data scientist to law professor illustrates the value of interdisciplinary thinking in addressing legal technology challenges

• The Massachusetts drug lab scandal, in which chemist Annie Dookhan falsified thousands of drug tests, became a formative moment that revealed how the legal system’s data practices were woefully inadequate

• Suffolk’s LIT Lab focuses on public interest law and access to justice, building tools that serve real people facing real legal challenges

• David draws a sharp distinction between “access to justice” and “access to an attorney”; technology can expand legal access in ways that don’t always require a lawyer

• Students at the LIT Lab build tools that go out into the world and serve real people, meaning their education creates direct public benefit

• The future of legal employment isn’t just about big law; the majority of attorneys in the U.S. work in smaller, solo, or public interest settings, and those contexts may benefit most from AI

• Work-life balance for David means choosing an intentional life: watercolor painting, poetry, walks, and a deep commitment to the examined life, values he absorbed from living abroad in Scotland

Notable Quotes:

1. “I sort of walked in. I was like, you don’t really know who I am... I’m going to ask you first off to just sit down and we’re going to spend 20 minutes doing doc review.” - David Colarusso (00:03:09-00:03:18)

2. “Almost all of them fell victim to something. It’s called automation bias. And so this was a way that I was able to sort of teach them about that, but without having to just say it to them.” - David Colarusso (00:04:48-00:04:55)

3. “They learned to trust the tool, and then the regime in which the tool performed the way they thought it did shifted. And they did not reevaluate their assessment of the tool. And so became over-reliant on it to a detriment to their accuracy.” - David Colarusso (00:08:23-00:08:37)

4. “The problem with a lot of these tools is they sort of upset our traditional signals for what represents quality.” - David Colarusso (00:11:02-00:11:08)

5. “I thought I was simplifying my life... And so that ended up with me becoming the data scientist at the Public Defenders here in Massachusetts.” - David Colarusso (00:32:11-00:32:25)

6. “I don’t want to make the mistake of thinking that access to justice means access to an attorney. In fact, some of the tools we make recognize that that’s not always the case.” - David Colarusso (00:34:57-00:35:05)

7. “I went from nature’s laws to people’s laws. And one of the things that’s nice about people’s laws is unlike nature’s laws... our laws are not set in stone. They’re not the weather. We can do something about them. They’re aspirational.” - David Colarusso (00:39:27-00:39:46)

8. “It mostly is my students, right? I’ve been a teacher in many different guises, and being able to interact with people and help them to sort of lead that examined life and to question what it is they want to do and how it is they want to look at the world and give them a tool set they can use when they go out into the world to make the impact they want to make.” - David Colarusso (00:38:34-00:38:55)

Clips

I Built My Own Major

The “Jagged Edge” of AI

You Aren’t Immune to AI Bias

Growing Up With Missile Tests

David’s classroom simulation on automation bias is more than a clever teaching exercise, it is a microcosm of one of the most pressing challenges facing the legal profession today. When students learned to trust an AI tool that was perfectly reliable, they stopped verifying its outputs. When that tool then failed in subtle ways, they performed worse than they would have without any AI assistance at all. This pattern—trust, reliance, reduced vigilance, degraded performance—is not unique to law students. It is the default human response to any decision-support tool, and it will play out in courtrooms, law offices, and regulatory agencies unless lawyers are deliberately trained to guard against it.

What makes David’s approach so compelling is that he doesn’t just warn students about these risks: he lets them live them. The pedagogical philosophy behind the LIT Lab is grounded in the idea that experiential learning changes people in ways that lectures simply cannot. When students build tools that go out into the world and serve real clients, they develop not just technical competence but moral accountability. That combination, technical literacy plus human judgment, is exactly what the legal profession needs most right now.

Closing Thoughts

Conversations like this one remind me of why I started LawDroid in the first place. David Colarusso represents something rare in legal education: a true polymath who has lived across multiple worlds (physics, film, public defense, data science, and now law school teaching) and brought the lessons of each one into his work with students. His automation bias simulation is one of the most elegant pieces of legal pedagogy I’ve come across, because it doesn’t just teach a concept. It makes students feel it.

What strikes me most is David’s underlying conviction that the law is aspirational. Unlike the laws of physics, which simply describe a universe we didn’t choose, our legal system is something we actively construct and can change. That perspective, hopeful, humanistic, and grounded in real public service, is exactly the kind of energy we need shaping the next generation of legal technologists.

For our Legal Rebels community, the lesson here is not to fear AI, but to approach it with clear eyes. Understanding where your tools work well, where they struggle, and when to question their outputs is not a technical skill, it’s a professional one. David is helping law students develop that skill before they step into practice. The rest of us need to keep developing it too.

If you found this article useful, you’ll love the LawDroid AI Conference 2026. April 28–29, virtual, and completely free — two days of keynotes, panels, and workshops on AI and the legal profession. I’d love to see you there.

Who chooses the problem we’re solving today?

You can build the most powerful search engine ever created, but someone still has to type in the query. You can build an AI that drafts a brilliant legal memo in seconds, but someone still has to decide that this memo, on this issue, for this client, is the thing that needs drafting right now, and not the forty-seven other things competing for attention.

We’ve spent the last three years getting very excited about AI’s ability to solve problems. We’ve spent almost no time asking who decides which problems get solved. And that silence tells you something about a profession that has, for too long, confused doing legal work with solving legal problems.

I call this the Selection Problem: the irreducibly human act of looking at the full landscape of possible problems and deciding which ones deserve our finite attention, resources, and care. It is, I believe, the domain where lawyers will create the most value in an AI-augmented world, and it is the domain that AI is least equipped to occupy.

If this sounds interesting to you, please read on…

This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.

The Objective Function

Here’s a concept from the world of algorithms that I think illuminates everything. When engineers build a system to find the “best” answer to a problem, they first have to define what “best” means. They write what’s called an objective function: a set of instructions that tells the system what to aim for. Minimize cost. Maximize speed. Find the shortest route. Each goal produces a different answer, and the system will faithfully pursue whichever goal you set. But, and this is the critical part, the system never chooses the goal for itself.

Want the cheapest solution? The system will find it. Want the fastest? It’ll find that too. But it will never look at your situation and say, “Actually, you’re optimizing for the wrong thing. Cost isn’t your real problem here.” That judgment, the act of deciding what matters, always comes from outside the system. Always.

A logistics AI doesn’t wake up one morning and conclude that carbon emissions matter more than delivery speed. A legal AI doesn’t decide that a client’s emotional wellbeing should outweigh the letter of the contract. Those are value judgments, and value judgments live upstream of computation.

Now, to be fair, AI has become remarkably good at creating structure from unstructured information. Hand it a pile of disorganized client documents, contradictory witness statements, or a sprawling regulatory landscape, and it will impose order. It will categorize, cluster, summarize, and surface patterns you might have missed. That is genuinely valuable, and it would be dishonest to pretend otherwise.

But structuring information is not the same as deciding which structures matter. AI can organize the mess on your desk into neat piles. It cannot tell you which pile represents the problem your client is afraid to name, or which pile contains the seeds of a crisis that hasn’t yet become visible. It doesn’t know what counts as a problem because “what counts as a problem” is a function of human values and worry, fear, sleepless nights, shame, but also institutional priorities, and contextual judgment, not data patterns.

The gap is between structured and what’s worth solving.

The Myth of the Liberated Lawyer

You’ve heard the pitch a hundred times. AI will handle the drudge work, the document review, the research, the first drafts. You’ll be free to “think more strategically.” You’ll “practice at the top of your license.” The tedious stuff drops away, and what remains is the good stuff: the high-level judgment, the creative strategy, the work you went to law school to do.

I want to take this claim seriously, because I think it contains a genuine insight buried inside a significant blind spot.

The insight is real: AI does, in fact, liberate time. When a task that took forty hours takes four, something has to fill the remaining thirty-six. And the optimistic case is that lawyers fill it with higher-order thinking.

The blind spot is: “think more strategically” is not a self-executing instruction. Strategy doesn’t simply appear when you clear calendar space for it. Strategy begins with problem identification, the act of surveying a complex, ambiguous, often contradictory landscape and deciding what the actual problem is. Not the obvious problem. Not the problem the client thinks they have. The real one.

And here is where the profession needs to have an honest conversation with itself. For too long, we’ve defined lawyers by their outputs: wills drafted, contracts reviewed, motions filed. We’ve treated legal work as a series of tasks to be completed, turning lawyers into, frankly, task rabbits. But the definition of a lawyer has never been “a person who produces legal documents.” A lawyer is a person who identifies and solves legal problems. The document is an artifact of the solution; it is not the solution itself.

When we reduce lawyering to task completion, we hand AI its easiest possible victory. Of course a machine can draft a will. The question was never whether it could produce a document. The question is whether it can sit across from a grieving widow, understand the family dynamics she’s too proud to articulate, and recognize that the real problem isn’t the will at all; it’s the estranged son and the business succession plan nobody wants to discuss. That’s problem identification. That’s lawyering.

Consider a general counsel with fifty matters on her desk. AI can help with every single one of them. It can draft motions, summarize contracts, flag regulatory changes, analyze discovery. But it cannot tell her which five of those fifty matters actually threaten the company’s strategic position. It cannot weigh the CEO’s appetite for risk against the board’s tolerance for ambiguity against the competitive dynamics of a shifting market. It cannot feel the political undercurrent in the organization that makes one seemingly minor compliance issue a powder keg and another a non-event.

Those are selection problems. And they require a human being standing in the middle of the mess, accountable for the consequences.

The Multiplication Paradox

Here is where the Selection Problem gets harder, not easier, with better AI.

Consider what happens when AI gets dramatically better at solving problems. Every task that used to take a team of associates a week now takes an afternoon. Contract review that consumed months of junior lawyer time happens in hours. Regulatory analysis across twelve jurisdictions, something that would have been a major staffing decision, becomes a Tuesday morning prompt.

This sounds like liberation. It is, in fact, a multiplication of the Selection Problem.

When your capacity to solve problems expands by an order of magnitude, the number of problems you could solve expands with it. That general counsel who had fifty matters on her desk? Now she can meaningfully act on all fifty. But she still has the same number of hours, the same budget, and the same board with the same risk tolerance. AI didn’t reduce the number of decisions she has to make; it increased them. Every problem that was previously too expensive to touch is now within reach, which means every one of them demands a selection decision that didn’t exist before.

A law firm that once had to triage aggressively because capacity was scarce now faces a different kind of scarcity: the judgment to know which of its newly solvable problems actually deserve solving. A legal aid organization that can suddenly process ten times the inquiries now confronts a question it used to answer by default through resource constraints: which of these clients’ problems do we prioritize when we can technically help all of them?

More capability, more choices. More choices, more consequential selection. The resource allocation question, who gets our finite attention, our finite hours, our finite best thinking, doesn’t get answered by better AI. It gets amplified by it.

And the counterargument, that AI will eventually learn to select problems through preference learning or value alignment, misses the point. Even if an AI could perfectly model a firm’s stated values and a client’s expressed preferences, it would still be optimizing against an objective function that someone had to define. The recursive problem remains: who decides what the AI should value? Who writes the function that tells the machine which problems are worth its attention? That’s still a human standing in the gap, making a judgment call, bearing the consequences.

The Selection Problem doesn’t shrink as AI improves. It grows.

The Exposed Gap

If the Selection Problem is as important as I’ve argued, you would expect the legal profession to have spent decades cultivating it. Training it. Rewarding it. Building institutions around it.

We haven’t.

Here’s what actually happened: AI didn’t create the Selection Problem. It exposed a gap that was already there. For decades, the economics of legal practice allowed us to avoid confronting it. When the billable hour was king, there was no incentive to ask whether a problem was worth solving; there was only an incentive to solve it and bill for the time. When task volume was the measure of a practice, selection was a luxury. You did the work in front of you. You didn’t ask whether it was the right work.

Law schools don’t teach problem selection. They teach issue spotting, which is not the same thing. Issue spotting is a bounded exercise: here is a fact pattern, find the legal issues. Problem selection is unbounded and human-centered: here is a client, a community, an institution embedded in a web of human relationships and competing pressures. What, among everything that could be a legal problem here, actually is one? And among those, which ones matter most?

The profession never built this muscle at scale because the old economic model didn’t require it. Task completion was profitable. Document production was measurable. Problem selection was invisible, done informally by senior partners with good instincts, never codified, rarely taught, impossible to bill for directly.

Now AI is stripping away the task layer. The work that used to fill our days and justify our billing is increasingly handled by machines. And what’s left, what AI cannot do, is the Selection Problem. The thing we should have been training for all along.

This is not a story about AI’s limitations. This is a story about ours. And it leaves us with a single, direct challenge: when the task work disappears, what remains underneath it?

Closing Thoughts

I started with a simple question: who decides what is worth solving? The answer, it turns out, is more revealing than I expected.

AI is the most powerful problem-solving engine humanity has ever built. It can draft, research, analyze, and synthesize faster and more reliably than any team of lawyers. And as it gets better, it doesn’t solve the Selection Problem; it makes the Selection Problem bigger, more urgent, more consequential. Every new capability is another fork in the road that requires a human being to choose a direction.

The profession’s future doesn’t belong to the lawyers who learn to use AI most efficiently. It belongs to the ones who can stand in a room full of newly solvable problems and say: this is the one that matters. Not that one. Not those. This one. And here’s why, and I’ll own the consequences of being wrong.

We’ve been task rabbits long enough. The machines are here for the tasks.

It’s time to choose wisely.

If you found this article useful, you’ll love the LawDroid AI Conference 2026. April 28–29, virtual, and completely free — two days of keynotes, panels, and workshops on AI and the legal profession. I’d love to see you there.

Register Now!

If you want to understand how a bias toward action — not endless deliberation — is what actually moves the needle on access to justice, you need to listen to this episode. Sam is at the forefront of legal innovation and brings a rare combination of practitioner experience, entrepreneurial grit, and systems-level thinking to the challenge of making justice work better for everyone.

From Courtroom to Code: How One Lawyer Decided to Build the Tools Justice Needed

Join me as I interview Sam Harden, Innovation Strategist at Affinity Consulting Group.

In this insightful podcast episode, Sam shares his winding path from Birmingham, Alabama to the forefront of legal technology — from teaching himself Python to build a court date search tool, to lobbying for criminal justice data reform, to helping legal aid organizations and law firms harness the power of AI. He dives deep into what he calls “Law Town,” a provocative vision of a future where lawyers supervise AI agents handling routine tasks so they can focus on what actually matters: counseling, advocacy, and human connection.

His stories and insights underscore a powerful philosophy — that the justice system is an ecosystem, and that small, well-placed interventions can ripple outward in ways that change lives. This episode is a must-listen for anyone who believes technology has a role to play in closing the justice gap, and wants a pragmatic, honest take on where AI can help and where the hard work still falls to us.

The Skinny

Sam Harden, Innovation Strategist at Affinity Consulting Group, brings a rare résumé to the legal tech world: criminal defense lawyer, self-taught coder, legal analytics entrepreneur, access-to-justice advocate, lobbyist, and now consultant to law firms and legal aid organizations navigating the AI era. Raised in Birmingham, Alabama by a family of pharmacists, Sam found his way to law through a passion for advocacy and a desire to push for justice in whatever form that took. His pivot away from trial law came not from a lack of passion but from burnout — and led him to build courtdatesearch.com, a tool designed to prevent the cascading spiral that a single missed court date can set off for vulnerable defendants. Throughout the conversation, Sam articulates a clear-eyed view of AI’s potential: it has made many things cheap and fast, but the hard problems — organizational culture, siloed law firms, human capacity constraints at legal aid organizations — remain stubbornly hard. His “Law Town” concept imagines lawyers as supervisors of their own AI workforce, freeing them to deliver the counseling and advocacy that no algorithm can replicate. Sam’s north star is simple: if the justice system is better ten years from now, it was worth it.

Key Takeaways:

• Sam’s “bias toward action” philosophy — being on “Team Do Something” rather than “Team Commission a Study” — has driven every chapter of his career, from building his first legal tech tool to advocating for criminal justice data reform

• A single missed court date can trigger a devastating spiral: warrant issued, arrest, job loss, eviction. Sam built courtdatesearch.com specifically to interrupt that chain of consequences for vulnerable defendants

• The justice system functions like an ecosystem, where small, targeted interventions, like reintroducing wolves to Yellowstone, can produce outsized, lasting change

• Legal aid organizations are not resource-limited in motivation or talent; they are human-capacity limited, and AI represents a genuine opportunity to extend their reach

• AI has made many legal tasks fast and cheap, but the hard organizational problems, siloed law firms, broken workflows, institutional resistance, are not solved by flipping the AI switch

• Sam’s “Law Town” concept envisions lawyers as supervisors of AI agent teams, offloading rote tasks like document review so they can focus on counseling, advocacy, and client relationships

• The greatest risk of AI in law is that lawyers become rubber-stampers, automating the drudgery without reclaiming the work that actually matters

• Work-life balance in the AI era is genuinely difficult; keeping up with the pace of change often means sleepless nights, and Sam is refreshingly honest about that struggle

Notable Quotes:

1. “If a problem is really bad and everybody agrees that the problem is bad, the best thing to do is do something. I’m on team do something. I’m not on team let’s talk about it, let’s commission a study and then write a report and then not do anything.” - Sam Harden (04:28-04:51)

2. “What if there was an easy way for people to just put in their name or their case number and check their upcoming court date? People are getting mailed a summons to court. A lot of people in the criminal justice system don’t have a fixed address.” - Sam Harden (20:52-21:12)

3. “You may start out with a driving while license suspended case. You don’t show up for your hearing, so the judge issues a warrant. And if you get pulled over, you go to jail. And while you’re waiting in jail, the job that you were working is going, where is this person?” - Sam Harden (23:05-23:35)

4. “I think the justice system is kind of like an ecosystem where small changes can have big impact years later. That’s kind of where I hope I have an impact — figuring out where I can push and where I can pull in the justice ecosystem to make it better.” - Sam Harden (46:07-46:29)

5. “AI has made a lot of things cheap. The hard things still are hard. You can automate a ton of stuff now, but figuring out how the pieces fit together is still really hard. And it takes time, it takes effort, and it takes experimentation, frankly.” - Sam Harden (34:53-37:49)

6. “The practice of law is the counseling, is the advocacy, is the handholding, is the talking to the client, making the client understand their options, being that personal helper for people.” - Sam Harden (39:39-39:54)

7. “Could a lawyer take all of the things that they don’t want to do — the rote document review — and have an AI agent run their playbook? And then the lawyer takes that and goes to the client and says, here’s how I would counsel you. And instead of charging by the hour for that, maybe it’s a different kind of value delivery.” - Sam Harden (40:07-41:09)

8. “The thing that makes it worth it to me is feeling like I’m having some sort of impact on the delivery of justice. If the justice system is better ten years from now, I’m happy.” - Sam Harden (45:04-46:35)

Clips

AI’s Hallucinated Case Law

AI Won’t Fix Lawyer Fiefdoms

Bias Towards Action

AI Can Draft Court Pleadings Now

Sam Harden’s career is a case study in what it looks like to take justice personally — not just as a professional obligation, but as a design challenge. His court date search tool didn’t change the law; it changed the information flow, and that was enough to interrupt a cycle of harm for people who had no other lifeline. That is systems thinking applied with humanity, and it is exactly the kind of ingenuity the legal sector needs more of.

What I find most compelling about Sam’s perspective is his refusal to be either a techno-utopian or a skeptic. He has used Claude Code, tested Google’s agentic tools, worked inside legal aid organizations, and consulted with law firms — and his conclusion is nuanced: AI lowers the floor on what’s possible, but raises the stakes for getting the organizational and human factors right. The “Law Town” vision is not a fantasy of lawyers replaced by robots. It is a vision of lawyers finally freed to do the work that only humans can do.

Closing Thoughts

Conversations like this one remind me why I started LawDroid in the first place. Sam Harden is not a theorist, he is someone who saw a problem, taught himself to code, and built something. The fact that courtdatesearch.com made zero dollars and still found its way onto the county clerk’s own website is exactly the kind of story I want our Legal Rebels community to carry with them. Impact doesn’t always look like a unicorn startup.

What strikes me most about Sam is his clarity about where AI fits and where it doesn’t. He is not selling a silver bullet. He is doing the hard, unglamorous work of helping organizations figure out how to actually use these tools, not just adopt them in name. That distinction matters enormously as our profession navigates what may be the most consequential technological shift in the history of law.

The “Law Town” framework is one I think every lawyer should sit with. Ask yourself: what are the things in my practice that I wish I never had to do again? Those are your candidates for automation. Now ask: what are the things I do that no AI can replicate — the moments where a client needs a human being who understands their situation, their fear, their options? That is your value. Protect it. Build toward it.

Sam’s north star — a justice system that is measurably better a decade from now — is one I share. I believe technology is one of the most powerful levers we have to get there. But as Sam makes clear, the lever only works if the people pulling it are thoughtful, honest, and relentlessly biased toward action.

In this month’s AI Double Take, LawDroid CEO Tom Martin and Chief Legal Futurist Sateesh Nori tackle a packed April in AI, from a surprising generational divide in attitudes toward AI, to the accidental leak of Anthropic’s Claude Code codebase, Google’s powerful open-source Gemma 4 release, and the rise of the first AI-powered solo billion-dollar company. The hosts debate whether AI should be regulated as a public utility, what the telehealth startup model could mean for legal access to justice, and why, despite the turbulence, both remain convinced the best is still ahead.

Key Takeaways

1. The Generational Divide — Gen Z vs. Gen X on AI

Counter to expectations, it’s Gen Z — not older generations — who are most resistant to AI in the workplace. Having grown up in a surveillance state, experienced social media’s harms firsthand, and come of age amid constant digital scrutiny, Gen Z brings deep skepticism to new tech. Gen X, by contrast, remembers the “before” — typewriters, microfiche, physical courthouse trips — and sees AI as liberation. The takeaway: the same technology looks entirely different depending on your “before snapshot.”

2. The Claude Code Leak — A Wake-Up Call

Around April Fool’s Day, Anthropic accidentally leaked the Claude Code codebase — including what appeared to be a pre-release model called “Mythos.” Key observations: (1) it can happen to anyone, even a $30B company; (2) the underlying system prompt code was simpler than expected — basic behavioral directives; (3) some instructions told the model to avoid leaving “fingerprints” when crawling for information, raising copyright questions; (4) Anthropic had apparently seeded the codebase with misleading decoy information before the leak. Once out, it spread instantly — the genie couldn’t be put back in the bottle.

3. Google’s Gemma 4 — Open Source Raises the Stakes

Google released Gemma 4, a powerful open-source model under Apache 2.0 licensing — meaning it can be freely copied, modified, and even resold. This puts real pressure on the defensibility of OpenAI’s and Anthropic’s proprietary model businesses, and dramatically expands what developers can build independently.

4. AI as Public Utility — The “Department of Intelligence” Idea

The Claude Code leak triggered a broader debate: should AI be regulated like electricity or water? Sateesh argued for a publicly regulated AI baseline — universally accessible, consistently priced — with private innovation building on top. Tom framed it as a “Department of Intelligence” or public library model: shared intelligence infrastructure that anyone can tap. Both hosts see self-regulation through market competition as insufficient.

5. The Two-Person Billion-Dollar Telehealth Company

A college dropout and his brother built a telehealth company — powered by AI and focused on GLP-1 weight loss drugs — to a $400M first-year revenue and billion-dollar valuation (verified by the New York Times). The model: AI handles scale, humans manage the customer relationship. The question for legal: why can’t this model be replicated for access to justice?

6. The Access to Justice Opportunity — Rethinking the Nonprofit Model

Sateesh challenged the traditional nonprofit legal model, noting that many legal aid organizations function more as jobs programs than delivery systems. With 92% of legal needs going unmet, AI-empowered individuals could scale their impact 10x beyond what a bureaucratic organization can achieve. LawDroid is actively building tools to enable exactly this kind of leverage.

7. Human Judgment + AI = Exponential Impact for Good

Both hosts’ final takes converge on optimism: the telehealth story proves that a single motivated person with practical intelligence and AI tools can create extraordinary impact. The challenge — and the mission — is to point that power toward good. Tom noted the current geopolitical climate as a prerequisite: nothing else can fully flourish until conflict is resolved.

Show Notes

Topics Covered

• Generational divide: Gen Z skepticism vs. Gen X techno-optimism toward AI

• Social media’s long-term impact on Gen Z’s mental health and trust in tech

• Personal anecdote: Tom’s Pixar-style AI photo and his daughter’s reaction

• The accidental Anthropic / Claude Code codebase leak (circa April 1, 2026)

• Leaked reference to a new Anthropic model: “Mythos”

• System prompt simplicity and “no fingerprints” crawling instructions

• Anthropic’s decoy/trap content pre-planted in the codebase

• Google Gemma 4: open-source, Apache 2.0, strong performance

• Competitive defensibility of proprietary AI models

• AI as commodity/utility — the electricity and internet analogies

• Proposal for a publicly regulated AI baseline (”Department of Intelligence”)

• First AI-powered one-person billion-dollar company (telehealth / GLP-1)

• Nonprofit legal aid model critique — the 92% unmet legal need figure

• LawDroid’s mission to empower AI-enabled legal access at scale

People & Organizations Mentioned

• Tom Martin — CEO & Founder, LawDroid

• Sateesh Nori — Chief Legal Futurist, LawDroid

• Ron Flagg — President, Legal Services Corporation (LSC); conference keynote

• Bridget McCormick — Conference keynote speaker

• Nikki Shaver — Conference speaker / thought leader

• Anthropic — AI company; Claude Code leak, “Mythos” pre-release

• Google — Released Gemma 4 (open-source, Apache 2.0)

• OpenAI / ChatGPT — Referenced in competitive defensibility discussion

• Unnamed telehealth founder — College dropout; first AI-powered one-person billion-dollar company (GLP-1 / weight loss drugs, verified by NYT)

Upcoming: LawDroid AI Conference 2026

• Dates: April 28–29, 2026

• Format: Virtual (attend from anywhere)

• Cost: Free

• Theme: The Year to Build

• Keynote speakers: Bridget McCormick (AAA), Ron Flagg (LSC), Nikki Shaver (LegalTech Hub), and more

• MC & Day 2 speaker: Sateesh Nori

• Register at lawdroidaioconference.com

Final Takes

Sateesh Nori:

“We’re already in April 2026, and I still feel like we haven’t crested the mountaintop on what’s coming. I’m with bated breath about what could happen tomorrow, next week, in May and June and beyond — not just in world politics, but in AI and the way our world is going to change, hopefully for the better.”

Tom Martin:

“I really hope the conflict going on right now resolves itself — nothing else can fully happen without that. But assuming it does, knock on wood: we’re at a place where everything seems possible. The telehealth story shows that someone who’s a college dropout can use the intelligence they have, with the aid of AI, to have an amazing impact. If only that were used for good — and I believe it can be — there would be so much more good in the world.”

AI Double Take is produced by LawDroid | lawdroid.com

If you want to understand how human-centered design and conscious self-awareness can help lawyers navigate the AI transformation without losing themselves in the process, you need to listen to this episode. Mia is at the forefront of legal design and AI adoption and brings a uniquely grounded, human-first perspective to one of the most urgent conversations in our profession.

Designing a Human Future for Law in the Age of AI

Join me as I interview Mia Ihamuotila, Legal Tech & Design Lawyer at Castrén & Snellman and Chair at the Legal Design Summit.

In this insightful podcast episode, Mia shares her journey from a childhood rooted in dance and creativity near Helsinki, through her studies in law and legal design, to building a hybrid role that bridges legal practice, design thinking, and AI strategy. She dives deep into her framework for the human-machine symbiosis, arguing that the real risk of AI is lack of intentionality in how we engage with it.

Her stories and insights underscore the importance of self-awareness, process thinking, and mindset in successfully navigating legal transformation. This episode is a must-watch for any lawyer curious about what it truly means to work alongside AI, not just how to use the tools, but how to remain grounded, creative, and deeply human while doing so.

The Skinny

Mia Ihamuotila, Legal Tech & Design Lawyer at Castrén & Snellman and Chair at the Legal Design Summit, shares her remarkable journey from a childhood spent dancing near Helsinki to building one of the most thoughtful and human-centered approaches to legal AI in the profession today. Raised outside the city with deep ties to nature, dance, and creativity, Mia was drawn to law for its power to shape society and its potential for transformation. After early traineeships in legal design with mentor Antti and a formative exchange year studying legal tech in Hong Kong, she wrote her master’s thesis at her current firm, empirically testing how legal design could improve the transparency and understandability of privacy policies. What sets Mia apart is her conviction that the human-machine relationship is not adversarial but symbiotic: that AI has the potential to elevate human thinking if we engage with it consciously, with self-awareness, and through multi-step collaborative processes. Her work now includes AI workshops and transformation consulting for both her firm’s lawyers and their clients, with the Legal Design Summit serving as a global platform to expand this conversation.

Key Takeaways:

• Mia’s path to legal innovation was shaped by a lifelong commitment to dance, creativity, and human connection: qualities she now brings directly into her legal design and AI strategy work

• Her origin story includes early burnout and a period of depression that led her to self-awareness practices like breathwork and meditation, which now form the foundation of her approach to balance and conscious design

• She argues for a symbiosis between human depth and machine breadth, when combined, these two dimensions create an entirely new dimension of capability

• The real risk of AI is not the technology but unconsciousness: the danger arises when people engage with AI without awareness, intention, or a structured process

• Practical AI adoption starts with mindset, not tools, mapping the lawyer skill set to include creativity, data literacy, interdisciplinarity, and design thinking before moving into tool use

• Mia conducts AI and design workshops for both internal lawyers and clients, framing this as a shared transformation journey rather than a one-directional education

• Legal design and legal tech are not separate disciplines: they are complementary life forces that belong under the broader umbrella of legal innovation

• The Legal Design Summit is evolving toward a chapter-based model with country-level ambassadors to maintain a coherent, year-round global community

• Balance is not just work vs. rest, it requires designing your environment around your nervous system’s needs, knowing when you need solitude, creativity, or social engagement

• Mia begins every day with breathwork, meditation, and movement, refusing to engage with the world before first centering herself

Notable Quotes:

1. “I usually talk about the symbiosis or the synthesis between the human and the machine and how it becomes kind of a wavy flow between those two. So it’s not about like man versus machine.” - Mia Ihamuotila (27:34-27:45)

2. “The machine has enormous capability to create breadth... but then as a human we have an immense capability to create depth in things. And I think these two dimensions, when combined, create a new dimension.” - Mia Ihamuotila (27:48-28:18)

3. “I really honestly believe that AI has the potential to elevate our thinking as human beings if we allow it to do so. And I think the tool for that is self-awareness.” - Mia Ihamuotila (28:23-28:36)

4. “I see this as a conscious evolution that is happening.” - Mia Ihamuotila (29:01-29:04)

5. “I don’t see the risks and the fear in there. I think the risk is this unconsciousness—the lack of consciousness in people in that case.” - Mia Ihamuotila (30:56-31:07)

6. “Everything starts from this mindset and the attitude and the mapping of skills, the lawyer skill set.” - Mia Ihamuotila (33:17-33:24)

7. “I don’t want to jump into the waves of the world before saying hi to myself.” - Mia Ihamuotila (44:08-44:14)

8. “I get to do transformation as the main driver in my career. And that’s been the main driver of my personal life forever. So that’s perfect.” - Mia Ihamuotila (46:21-46:35)

Clips

Death, Rebirth, and New Beginnings

Design and Tech: Inseparable Forces

Human-AI Symbiosis

Use Process Thinking Without AI

Mia’s story is a reminder that the legal profession’s transformation is not just technological, it is deeply personal. Her journey from dance studios and burnout to building a hybrid law career grounded in design thinking and conscious AI adoption reflects the kind of inner work that most transformation conversations leave out entirely. She brings to the table something rare: a framework for remaining human not despite the AI revolution, but through it.

What makes her perspective especially compelling is the practical dimension. It’s not enough to talk about consciousness and symbiosis in the abstract. Mia is in law firms, running workshops, mapping skills, and guiding lawyers and clients through the early and middle stages of this transformation, meeting people where they are and building the capacity they need to move forward. Her work at the Legal Design Summit amplifies this further, building a global community of practice that embodies the very transformation it champions.

Closing Thoughts

Talking with Mia was genuinely one of those conversations that changes how I think about what we’re doing in this space. As someone who has been building legal technology for years, I sometimes get caught up in the tools: the capabilities, the models, the use cases. Mia reminded me that the foundation of all of that has to be the human being holding it.

Her concept of conscious engagement with AI is something I want every lawyer in our community to hear. The technology is not the problem and it is not the solution. The quality of our attention, intention, and self-awareness is what determines what we build with it and what it does to us. That’s a profound reframe.

What inspires me most is the way Mia has integrated who she is, dancer, designer, lawyer, meditator, into a coherent whole rather than treating these as separate identities. That kind of integration is, I think, exactly what the profession needs right now. Not lawyers who bolt AI onto existing practice, but lawyers who are willing to grow into a fuller version of themselves and let their work reflect that.

For our Legal Rebels community, Mia’s story is both a challenge and an invitation. The invitation is to approach this transformation consciously, to start with mindset before tools, to design your environment around your nervous system, and to begin the day by saying hi to yourself before jumping into the waves of the world. The challenge is to do that inner work consistently, even when the pace of change makes it tempting to skip it.

The future of law belongs to those who can hold both depth and breadth, who can be fully human while working alongside machines that are becoming more capable by the day. Mia is living proof that this is not only possible but extraordinarily exciting.

We’ve all heard the pitch. Staring at a blank page? Just prompt AI!

In seconds you’ve got a working draft: structured, coherent, surprisingly competent. The blank page problem, that ancient enemy of productivity, has been vanquished.

Except the blank page was never just an obstacle. It was a space of possibility, of roads untraveled. And for lawyers, it was the space where the most important part of our work actually happens.

I call it the First Draft Trap!

The moment you accept an AI-generated draft as your starting point, you’ve already made the most consequential decision of the entire project; and you made it by not making it. You let the machine choose your direction, your framing, your theory. Everything that follows is editing. And editing, no matter how rigorous, is not the same as thinking.

If this sounds interesting to you, read on…

This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.

The Cognitive Hijack

There’s solid psychology behind why this happens. Daniel Kahneman and Amos Tversky demonstrated in their landmark 1974 paper, “Judgment under Uncertainty: Heuristics and Biases,” that once people are exposed to an anchor, even a transparently arbitrary one, it distorts their subsequent judgments. In their experiments, subjects who watched a roulette wheel spin to a random number still let that number influence their estimates of completely unrelated quantities. The anchor held even when people knew it was meaningless.

An AI first draft is the most seductive anchor imaginable. It’s not random; it’s plausible. It’s well-organized. It sounds like something a lawyer would write. And that’s precisely what makes it dangerous. You know intellectually that it’s just one of many possible approaches. But the anchor holds anyway. Kahneman would tell you that knowing about the bias doesn’t protect you from it.

There’s an even more precise concept here. In 1942, the psychologist Abraham Luchins identified what he called the Einstellung effect: once a person finds a solution that works, they become functionally blind to better solutions, even when those alternatives are obvious to someone encountering the problem fresh. The first working solution doesn’t merely compete with alternatives. It blocks perception of them entirely.

That’s the First Draft Trap at the pyschological level. The AI draft isn’t just one option you happen to prefer. It’s a cognitive filter that prevents you from seeing the other options that were available to you: the roads you never even noticed you didn’t take.

The Anti-Socratic Method

Now consider what this means for a profession built on the opposite instinct.

From the first day of law school, we’re trained to resist the obvious answer. The Socratic method exists for exactly this reason. A good professor hears your confident response and asks: What else? What if the facts were different? What’s the argument on the other side? The goal isn’t to arrive at an answer. It’s to build the mental habit of holding multiple possibilities in tension before committing to any of them.

The First Draft Trap is the anti-Socratic method. It delivers a confident answer before you’ve even formulated the question properly. And instead of interrogating it, you polish it.

Think about what a senior partner actually does when a junior associate brings a memo. The partner’s value isn’t better writing. It’s peripheral vision: the ability to see what the memo doesn’t address. The argument it didn’t consider. The framing that would land differently with this particular judge or this particular jury. That capacity to see beyond the document in front of you is what clients pay premium rates for. And it’s exactly the muscle that atrophies when your default workflow begins with “generate a draft.”

There’s a direct parallel to case strategy. A litigator who latches onto the first viable theory of the case is a dangerous litigator; dangerous to the client. The great ones hold three or four theories in suspension, stress-test each against the facts, and only then commit. That uncomfortable period of ambiguity, of genuinely not knowing which path is best, is where the real lawyering lives. The AI first draft is the equivalent of committing to a theory of the case before you’ve finished reviewing discovery.

System 1 Hijacks a System 2 Profession

Kahneman’s two-system framework gives us a clean way to describe what’s going wrong. System 1 is fast, intuitive, pattern-matching. System 2 is slow, deliberate, analytical. The practice of law, at its best, is a System 2 discipline. We’re trained to override our gut reactions, challenge assumptions, and think through consequences before acting.

The AI first draft feels like a System 2 output. It’s structured, footnoted, methodical. But your decision to accept it as a starting point is pure System 1: a fast, intuitive grab at the nearest plausible answer. You’ve used a sophisticated tool to bypass the sophisticated thinking the tool was supposed to support.

What to Do Instead

None of this means stop using AI. It means stop using it to skip the part that matters.

Before you ever ask for a draft, ask for the map. Try a prompt like this:

I’m working on [describe the matter, motion, or document]. Before drafting anything, give me:

1. Three fundamentally different strategic framings for this problem.

2. For each framing, the strongest argument in its favor and its most serious vulnerability.

3. Which framing best fits [the client’s goals / the audience / the procedural posture].

Do not write a draft. I want to choose the direction before we start building.

That last line is the key. It keeps you in the driver’s seat during the phase that matters most. You’re using AI to expand the possibility space before you collapse it, not after.

In Kahneman’s terms, use AI to fuel System 2, not to hand the controls to System 1. Let the machine generate options. You exercise judgment. And, also don’t forget to add your own ideas and explore them too!

The blank page was never your enemy.

It was the last place where all possibilities were still alive; where your judgment, your experience, and your capacity to see what others miss had room to operate. The First Draft Trap paves over that space with something that looks like progress but might be the most expensive shortcut you’ve ever taken.

For lawyers, the ability to see what isn’t there is the whole game.

Don’t let the first draft trap blind you to it!

If you found this article useful, you’ll love the LawDroid AI Conference 2026. April 28–29, virtual, and completely free — two days of keynotes, panels, and workshops on AI and the legal profession. I’d love to see you there.

Register Now