?? Let's Talk About AI Agents in Legal (And Why They Matter) [T&C's Volume 27]

Welcome to Volume 27 of T&C's and thank you for sticking around! In this edition, we'll be talking about AI agents, catching up on some paradoxes, and visiting Dubai with my team!

AI Agents & Agentic Tools 101: What They Are, Their Impact in Legal, and Some Cautions

Chances are you've heard someone talk about "Agentic AI" recently (including yours truly and my colleague Tanja Podinic ). Although some see them as transformative across the board, I think AI agents are an evolution— a welcome one — of how we've been approaching AI in legal and the way we do tasks.

What's an AI Agent? ???

To start, AI agents are semi-autonomous systems (at least in the legal world, where we have ethical and regulatory guidelines regarding this) that can make some decisions and take actions to achieve specified goals with limited human interaction. They're not just chatbots —although LLMs like GPT-4o and Claude are the core of these systems — it's a computer-based "assistant" that can dynamically control its own processes and make decisions about how to accomplish tasks, within a predefined set of conditions. Not only that, users can expect an AI agent to engage and use other tools and software, adapt to user expectations, and combine multiple capabilities (so a chatbot could be a subpart of an AI agent, but not viceversa).

If this sounds a lot like an AI-based workflow, you're not wrong! Agents can be somewhat autonomous systems or predefined implementations. When we talk about AI-based workflows and AI agents, the difference is whether there are predefined paths (a workflow) or the ability to dynamically make changes in the process to achieve the goal (agents). Even simpler than that might be workflows are a predictable set of rules and agents offer flexible model-driven decision-making.

Why are AI Agents experiencing a boom right now?

There's a good reason: AI technology is finally maturing enough to make them possible. As LLMs have matured, they've become able to reliably call other tools and software for assistance (and humans, of course), and they're promoting efficiency gains (for example, Wiley , a publisher, reported a 40% improvement in efficiency using Salesforce AI agents). Next, for lawyers specifically, there's a potential for hundreds of thousands of dollars in productivity improvements, with significant time savings at the same time. Finally, as enterprise tools like LeAh by ContractPodAi become more prevalent, with enterprise adoption, more and more businesses are adopting agentic solutions to solve complex problems.

First steps: where are agents making inroads in legal teams?

Here's where I see agents shaking things up:

1. Contract Analysis: Imagine an agent that doesn't just spot issues in contracts but actively works through complex documents, identifies potential risks, and suggests specific modifications based on your playbook.?
2. Legal Research: Rather than just searching case law, an agent could break down complex research tasks, explore multiple databases, synthesize findings, and present a structured analysis - all while maintaining a clear audit trail of its reasoning.
3. Due Diligence: Agents could autonomously work through massive data rooms, flagging issues, requesting clarifications, and organizing findings in real-time.

We're also going to see a big uptick in the use of AI agent for case analytics, arbitration management, and even compliance monitoring, each of which is showing improvements in accuracy and risk reduction through human-machine interaction.

Not All Sunshine and Rainbows, Though

But here there are two catches (because there's always at least one catch in legal tech ??)... Doing a good job with AI agents needs clear user guidelines and implementation requirements.

First, AI agents acting as tools can help legal professionals analyze documents, conduct research, and process information - but must always do so within the bounds of user permissions and oversight. This means having AI agents with (i) clear specification of agent capabilities and limitations, (ii) transparent logging of agent actions and reasoning, and (iii) robust user controls for reviewing and approving agent recommendations.

Second, legal teams implementing legal AI agents must ensure specific safeguards and guardrails, including:

• Start simple and validate each capability before adding complexity - errors in legal work compound quickly
• Maintain complete transparency in how the agent reaches its conclusions - every recommendation must have a clear, traceable rationale
• Design clear interfaces that specify exactly what the agent can and cannot do - just as we set boundaries with any legal technology

The goal isn't to build the most sophisticated system - it's to build the right system that empowers legal professionals while maintaining appropriate controls.

Paradoxes, Laws, Maxims, and More

I've been talking about Jevons' Paradox for several years now, particularly as legaltech and now AI for legal has become more prevalent. Satya Nadella brought it to the front of mind towards the end of January after posting about it online.

At the core, Jevon's Paradox is a Victorian-era application of the economic law of demand that states that when technological advancements make a resource more efficient to use, as the cost of using the resource drops, overall demand increases causing total resource consumption to rise. At the time, Jevons was talking about how improvements in coal production would drive increased consumption of coal (instead of decreased consumption as the Victorian government hoped).

But Nadella wasn't talking about it in a legal-specific case, he was talking about it generally as access to cheaper, faster, more efficient AI models such as DeepSeek R1 that require less resources to run — means that more people will be using AI. While some folks think that AI tools and agents will make lawyers go away, I'm thinking that by making legal services more accessible through technology, more people want legal services. That small business that couldn't afford manual contract review suddenly wants automated review of every vendor agreement. The reduced cost of obtaining legal assistance will, happily (at least for lawyers) mean more work on servicing clients and business partners (with less of the activities that lawyers don't like).

Check out the article I recently published on Jevons' Paradox, Parkinson's Law, and Illich's Law and let me know what you think!

ContractPodAi in DubAi: Hummus and LeAh For All!

Much of our ContractPodAi team was recently in Dubai for our annual kick-off, with amazing customers, partners, and new exciting releases all combining for a tremendous five-day experience.

I had the opportunity to ask our team about some of their favorite things from the recent trip (other than seeing me eating an endless supply of hummus):

Getting to see, interact, and learn how our products improve life for our customers in real ways was great! I learned that we aren't benefiting just legal, but stakeholders throughout the organization. Giving legal teams the data they've always needed means that business value of legal departments soars. - Shawn Blaney

My top three takeaways? (1) GCs and CLO's have a tremendous opportunity to enact change, but they need to advocate for themselves and their teams. (2) AI agents are going to help legal drive revenue and business, creating vast new possibilities. (3) Current geopolitical trends are going to require legal teams to truly have legal and contract visibility across borders, and technology can make them more nimble. - Andrew F.

Being able to see in-person that not only do we do great things here at ContractPodAi, but we also have great people. Our customers and partners are showing off what they can do with LeAh and our CLM across a range of products, and that is incredibly inspiring. And hearing Lidia Kamleh (from the DUBAI FUTURE FOUNDATION ) talk about how she is using our technology to recover time so she can work on more exciting projects was wonderful. - Maureen Perry

With that, I'll say ciao for now — although if you're looking for some photographs, please keep scrolling!

(Pictures courtesy of Sarah Amberin I. , Adam Savage , Naomi Gaston , and Jerry Levine )