The AI Agent Revolution: Key Insights from Tom Martin on the Future of Legal Tech
Where I offer a bite-size big picture view on the journey from chatbots to autonomous AI and the road ahead
I recently had the honor of being interviewed by Victor Li on the ABA Journal’s Legal Rebels podcast. I shared my insights about legaltech’s evolution from simple chatbots to sophisticated AI agents that could fundamentally transform how lawyers work. Having spent nine years building AI solutions for the legal industry, I offer a unique perspective on where we’ve been and where we’re headed.
If you’d like to listen to the the full podcast episode, just click ‘Play’ below.
If you’d like to skip the listen, and get the key takeaways, 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.
Key Takeaways
AI Agents vs. Generative AI: Understanding the Difference
“Agents use generative AI to accomplish what they need to… generative AI is what empowers agents to take multiple steps to accomplish goals. So they’re not two different things. In fact, they rely on each other.” - Tom Martin
The key distinction: While generative AI responds to prompts, AI agents can autonomously pursue complex goals with minimal supervision, creating their own plans and using multiple tools to achieve outcomes.
The Evolution of Legal Tech Workflows
I broke down the evolution of legal tech workflows, using eDiscovery as an example:
1. Traditional, human workflow: Human associates manually tag documents by reviewing them page by page in a warehouse populated with banker boxes of files and documents and hundreds of legal professionals. They may keep track of the document data they manually extract in an Excel spreadhseet.
2. Rule-based workflow: Documents are scanned and the text within them is recognized using optical character recognition (OCR). A computer program then uses keywords to identify specified legal topics. Regular expressions (or ‘regex’ for short) are used to find / match patterns and special characters, like tabs and spaces.
3. AI workflow: Documents are scanned and the text within them is recognized using optical character recognition (OCR). The text may be tokenized, vectorized, and stored within a vector database. AI models, like Claude Sonnet 4 or GPT-4o, are used to understand and tag content based on semantic meaning. The step-by-step process is programmatic.
4. Agentic workflow: An agentic workflow is presented with documents and the goal to identify documents containing certain legal topics. The agent creates a document analysis plan, then identifies what tools to use to accomplish the plan. The agent would use its tools to scan, recognize, and ingest the documents, then use AI models to identify topics and categorize them accordingly. The step-by-step process is not static, but generated on the fly.
5. True AI agents: True AI agents are rarely employed. A true agent would act autonomously to source documents in real-time as they become available, set its own goals and create document analysis plans to accomplish those goals. It would employ tools, dynamically adapt research plans across documents in response to discovered data, and work autonomously for hours on end without human intervention. The step-by-step process is dynamically responsive to changes and generated on the fly.
All of the aforementioned workflows require humans to remain in the loop to ensure quality, accuracy and reliability.
The Adoption Reality Check
“The future is here, it’s just not evenly distributed. It’s definitely true for law firms.”
According to recent reports:
86% of in-house counsel use generative AI weekly (Wolters Kluwer)
78% of law firms report weekly AI usage (Clio)
But the reality is that most lawyers are using AI for basic tasks like summarization and email drafting, not integrated workflows and much less for anything “agentic.”
Insights
The Cost of AI Agents
I warn about something many lawyers don’t think about with all the excitement about AI agents: their hidden expense.
“Each one of those calls and inference steps cost money. And so it could get very expensive, very quickly, especially… with multi-agent architecture.”
When a research agent orchestrates multiple sub-agents, each making numerous AI model API calls, the costs can spiral quickly, not to mention the amount of time expended by additional agentic complexity — crucial considerations for firms evaluating these tools.
Agents aren’t magic; they costs time and money.
So, it’s important to consider the need for AI agents. Are AI agents necessary and fit to purpose, or are they simply too cool to pass up?
I would advise any lawyer to consider what I call the ‘Intelligence Razor’: When choosing between technology solutions, select the least sophisticated option that adequately solves the problem.
It could be that, all things considered, a rule-based workflow is the quickest, least expensive and highest quality best choice (albeit the least sexy).
The Proactive Law Firm of the Future
AI will transform law from reactive to proactive:
“Our work as [legal] professionals has largely been reactive. Somebody comes with a lawsuit, or they wanna file a lawsuit… I think the pro-action, the ability for lawyers to actually have a much closer relationship with their client, will be enabled by generative AI.”
I envision law firms building real-time legal analysis into clients’ daily operations, creating “risk loops” that flag issues before they become problems—opening up entirely new business models for legal services.
The Drone Analogy: More Jobs, Not Fewer
Addressing the perennial fear of job displacement, I offer a comparison:
“When drones came onto the scene, people were really worried that the traditional (artisanal) pilot would go away… Actually, what happens, a drone doesn’t just fly itself. It actually requires a support crew.”
The parallel: AI won’t eliminate lawyers but will create new roles—legal AI data scientists, legal engineers, and positions we haven’t yet imagined.
Notable Facts
Martin’s prescient 2016 presentation showed AI reaching human-level intelligence by 2025, remarkably close to current expert predictions of 2-5 years
AI agents can now work autonomously for up to 8 hours (according to Anthropic’s latest reports)
Martin developed CiteCheckAI to combat hallucination problems in legal citations
2025 is being called “the year of agents,” but Martin believes legal will lag behind
Quotable Quotes
On AI becoming mainstream:
“All of this becomes much more interesting when it becomes boring, when it becomes just another tool.”
On human value in the AI age:
“At the end of the day, we carry malpractice insurance… I don’t know of any AI agent or model that’s backing up what they do with indemnity at this point.”
On responsible AI use:
“We all have to find a responsible way to work and live with AI, because it’s here to stay, and we need to have rules and principles that guide how we interact with it.”
On the access to justice gap:
“Most people can’t afford lawyers, and these types of tools allow organizations to amplify and multiply the number of people they can reach because it’s just not physically possible, even if we tried to throw human lawyers at the problem.”
What’s Next?
I provide an outlook for the future of AI agents and agentic workflows:
Near-term (1-2 years):
Law firms will develop AI policies and procedures
Basic agentic workflows will emerge in larger firms
Cost optimization will become crucial as multi-agent systems proliferate
Medium-term (3-5 years):
Proactive legal services will create new business models
Legal support roles will multiply and diversify
AI agents will handle routine legal analysis in real-time
The Bottom Line
My message is clear: The legal profession isn’t disappearing, it’s evolving. The firms that thrive will be those that embrace AI as a tool for transformation, not just efficiency. The key is focusing on outcomes, not just processes.
For lawyers worried about their future, I offer reassurance with a dose of reality: Your job description is changing, not disappearing. The question isn’t whether to adopt AI, but how to do it responsibly and strategically.
Want to learn more? Share your thoughts in the comments below.
By the way, did you know you that I now offer a daily AI news update? You get 5 🆕 news items and my take on what it all means, delivered to your inbox, every weekday.
Subscribe to the LawDroid AI Daily News and don’t miss tomorrow’s edition:
LawDroid AI Daily News, is here to keep you up to date on the latest news items and analysis about where AI is going, from a local and global perspective. Please share this edition with your friends and colleagues and remember to tell me what you think in the comments below.
If you’re an existing subscriber, you read the daily news here. I look forward to seeing you on the inside. ;)
Cheers,
Tom Martin
CEO and Founder, LawDroid