Preparing Legal Teams for an Agentic AI Future

In the summer of 1996, my sisters and I played with our new Gameboy Pockets in the backseat of our my dad’s car on a long road trip. As we ran around with Mario and organized our Tetris blocks, we said to each other: “You know what would be awesome? If this screen was in color.”

Naturally, we were quite sure that this was an entirely unique idea. But alas, we were elementary students with no ability to make this totally-unheard-of invention happen.

Two years later, the Gameboy Color hit stores everywhere and we were certain it had been our idea first. What torture it was to have missed out on that huge moneymaking opportunity simply because we were too young to make it happen ourselves. I still think about it sometimes.

There are two kinds of legal professionals in the age of artificial intelligence: those who see what’s coming and, lacking the skill set to embrace it, keep on keeping on and hope for the best; and those who see what’s coming, acknowledge they don’t yet have the skill set to embrace it, and then go ahead and build those skills.

If the last few years have taught us anything, it’s that:

1. AI is evolving fast. Faster than any one of us can keep up with. All you can do is get your priorities straight, find the right advisors, and use their guidance in your decision-making as you go along for the ride.
2. Legal AI shouldn’t replace humans; it should support them. This is a new iteration of the same people/process/technology trifecta you know so well; it’s just that the technology can talk now.
3. There is no more “if” when it comes to AI for legal. Transformation isn’t coming; it’s well underway. You don’t have to do everything at once, but if you don’t start doing something with it, you’re falling behind your competitors, the reality of your legal data challenges, and the expectations of your clients.

You probably don’t need any further convincing of any of this. What you need is some practical tools and advice for how to move forward.

What should you be doing with AI right now? What might you be doing tomorrow? What skills will you need to bridge those gaps—and how can you start building them?

McKinsey published a report late last year entitled Agents, Robots, and Us: Skill Partnerships in the Age of AI, and if you haven’t seen it yet, it’s a great place to start digging into these questions.

In particular, the chapter entitled “How AI Changes Skills” is a goldmine of insight for legal data professionals. Let’s dig in.

Why Legal Work Is Uniquely Positioned for Human and AI Collaboration

McKinsey’s research shows that while a significant percentage of work hours across industries are technically automatable, most roles won’t be eliminated. Instead, tasks within roles are being reshuffled.

“Each wave of technology has changed what workers do. The difference today is speed. Until 2023, the need for AI-related skills grew at roughly the same pace as for cloud computing, cybersecurity, and other digital skills,” the report states. “After the rise of generative AI, it accelerated sharply: Nearly 600 new skills appeared in job postings over the past two years—about one-third of the total added in the past decade—many of which are tied to AI and its enabling technologies.”

In legal work, we’re already seeing AI automate familiar, rote, data-intensive tasks that machines are well equipped to tackle:

• Routine document classification
• First-pass review
• Pattern recognition across massive data sets

Discerning and telling the story those data points represent, however, is a fundamentally human exercise. So is advising and representing a client based on the insights gathered. And now, knowing how to prompt and validate an AI-powered tool to uncover those insights faster is becoming a critical legal skill, too.

So, as AI takes on more of the repetitive cognitive load, human effort shifts toward those delightful intellectual exercises that drew most of us to the legal profession in the first place:

• Judgment, interrogation, and interpretation
• Strategy development
• Risk assessment
• Communication and advocacy

Legal data work is both highly structured (governed by known rules, standards, procedures) and deeply contextual (executed with a deep understanding of the facts, nuances, judgments, and consequences of a given matter). This combination makes it an ideal environment for human-in-the-loop AI, where systems handle scale and speed, and humans retain control over interpretation and outcomes.

In law, accuracy and defensibility matter a heck of a lot more than automation. But automation helps, too. For legal teams to embrace it, they must first build trust in AI via purposeful transparency, validation, and human oversight.

Which AI Skills are Most Important for Legal Professionals?

AI literacy is quickly becoming essential to legal competence.

McKinsey notes that demand for AI-related, non-technical fluency is rising faster than almost any other skill category, climbing 6.8 times from 2023 to 2025 across job categories.

For lawyers, competence has always evolved with technology. Just as email, digital documents, and text analytics became standard in our professional lives, AI literacy is on a similar trajectory.

That doesn’t mean every lawyer or legal technologist must use AI in production today. But it does set a bar for understanding:

• What AI can and can’t do
• How automation impacts the risks, proportionality, and accuracy of a given project
• How to evaluate legal AI tools responsibly

Those who invest in these skills early can help shape best practices for their teams and the broader industry.

McKinsey found that over 70 percent of skills that employers valued in 2025 are used in both automatable and non-automatable work. The key, then, will be to learn how to conduct that people/process/technology symphony in new and creative ways.

For legal professionals, that evolution could look something like this:

Thanks to AI, the future of legal practice has less emphasis on:

• Manual review as a primary value driver
• Memorization or brute-force analysis
• Linear, one-document-at-a-time workflows

Instead, AI-enabled legal teams will place more emphasis on:

• Framing the right questions for AI-assisted systems
• Evaluating outputs critically and defensibly
• Designing workflows that balance speed, accuracy, and risk
• Knowing when not to rely on automation
• Human-to-human collaboration and strategic discussion based on available data

In other words, you don’t need to build models or major in mathematics. AI fluency in the law, for many of us, is less about technical specialty and more about professional ability. Focus on understanding how AI reasons, where it fails, and how to validate its work.

Agentic AI and Building Legal Skills the Responsible Way

Already, legal AI platforms have extended beyond conversation and research and, with agentic AI, can perform specific tasks autonomously as directed by users. The skill shift required for legal practitioners becomes even more pronounced when we move into this world of action.

Agentic systems, in legal contexts, must be paired with control and require human oversight. Thus, an emerging skill set for legal professionals can’t focus only on setting up an agentic tool and letting it run free. Learning how to set clear constraints, review citations and reasoning from the AI, validate outputs, and when to intervene or alter a workflow are also essential.

That may all sound a bit fuzzy if you aren’t already familiar with agentic AI, but rest assured that the most effective learning doesn’t come from theory alone. Guided experimentation, thoughtful evaluation, and shared standards will give you the hands-on experience you need to level up.

This approach can include tactics like:

• Conducting controlled pilots. Talk to a trusted software or services provider for a look at what they offer, how humans stay in the loop, and even whether they can create something bespoke to suit your team’s go-to methods.
• Defining clear validation benchmarks. While building AI, and especially agentic AI, into your standard procedures may feel like changing your paradigms in a big way, it shouldn’t be a blind leap of faith. Rely on familiar validation metrics—like those you learned when you first started using TAR—and, again, talk to the experts about what goals to set and how to measure against them.
• Prioritizing transparent, reviewable AI behavior. Look for tools that provide some insight into how they operate, why they make each determination, and citations backing up their work. You can avoid black boxes, improve defensibility, and learn a lot about how AI thinks this way!

Remember that lagging behind won’t serve your clients—or your career—nearly as well as cautious and considered exploration will. So go ahead and test the waters; you’ll probably find them quite fine.

Looking Ahead

The McKinsey report’s message is refreshingly optimistic: the future of work belongs to the people who adapt their skills, not the robots who can take over some of them.

In legal data intelligence, the opportunity is especially clear: AI can scale while humans handle meaning. The professionals who thrive will be those who know how to orchestrate both.

If you’re interested in exploring what that looks like in practice—particularly how to think about agentic AI safely, defensibly, and with humans firmly in control—check out our Agentic AI Toolkit. It’s designed to help legal teams cut through the noise, build shared understanding, and develop the skills that matter most in this next phase of legal technology.

Graphics for this article were created by Kael Rose.

---