Key Takeaways from ILTACON 2026: From Groundwork to Proof

Last year’s ILTACON was about proving AI could deliver results. This year, in Nashville, the conversation moved a step further: do firms have the fundamentals in place to get real value from it?

That means data structured well enough to trust, people brought into the change rather than simply handed a new tool, technology matched to how work actually gets done, oversight of the agents doing that work, and evidence that it’s making a difference.

The gap between what AI can do and what firms have built around it was one of the clearest themes of the week.

1. From Naming Data to Connecting It

Last year, much of the conversation centered on search and finding the right precedent within a mountain of documents. This year, it went a level deeper into how a firm’s information is structured in the first place.

Heath Harris of NetDocuments laid out the efficiency case: a well-defined taxonomy can cut the context an LLM needs to consume by 40 to 50 percent, but only if a firm has first agreed on its own language. Is it a stock purchase agreement or a share purchase agreement, and who decides? His caution: “More is not better,” especially once you’re trying to get the most out of an LLM.

By day three, that thread had expanded into how firms connect information across systems and teams. Igor Labutov of Epiq gave a concrete example: “Procurement data, contract data, disputes, all these things are different groups, organizations. Prior to now, it was really hard to bring those together... That unlocks a whole new set of value propositions.”

That’s where ontology enters the conversation: moving beyond naming and organizing information to mapping how matters, clients, agreements, precedent, and other information relate to one another.

Next step: Audit how your firm’s data is structured today, not just where it lives. Look at whether your taxonomy simply identifies information or also captures the relationships needed to make that information useful.

2. Change Management Starts Before You Buy Anything

Firms have no shortage of strong AI products to choose from. The harder question is getting people to actually use them.

Harriet Joubert-Vaklyes of Michael Best & Friedrich LLP put it simply: “You need to start with questions long before you decide that a new tool is necessary.”

A live poll backed her up. Peer champions, rather than top-down mandates or training, were named the most effective way to drive adoption. Yet many firms acknowledged that their change management planning still happens shortly before lawyers are expected to start using something new.

That isn’t separate from how people feel about the changes happening around them. Sessions on resilience and overcoming fear, led by practitioners including Brad Kaufman of Greenberg Traurig and Elaine Dick of BakerHostetler, drew some of the week’s strongest attendance. People are being asked to change how they work, and the human side of that change deserves as much attention as the technology.

Next step: Start your change management plan before you shortlist vendors, not after you sign a contract. Identify peer champions early and give them a role in the decision.

3. Match the Tool to the Workflow

Two sessions challenged the idea that a single AI platform will necessarily serve the entire firm equally well. Different practice groups have different workflows, problems, and expectations for AI, making evaluation at the practice-group level increasingly important.

There’s a budget reality here, too. Joannie Murray of Krieg DeVault gave one of the week’s most candid descriptions of running IT at a mid-sized firm: “Law firm users expect instant access from everywhere, seamless product integrations, mobile first experiences, zero downtime, AI-assisted workflows, and the shiny new product no one knows anything about, but they want immediately. The challenge is where we invest finite resources to create the greatest user impact.”

Next step: Before standardizing on a platform, map the specific workflows and practice groups it needs to serve. A tool that works well for litigation support may not be the right fit for transactional work.

4. Agentic AI Needs Managers, Not Just Adopters

As agents take on more work, firms also need to think about who is responsible for what they do.

Tim Fox of Ogletree Deakins was direct about where people still fit in: “There’s not a workflow I’m aware of that shouldn’t have a human in the loop when it comes to AI.”

He also flagged a shift in how firms talk about value: “We’re switching from efficiency to enablement, talking more about what you can do with AI instead of just how much time is saved.”

Spending is following that shift. Fox noted that AI budgets, often estimated at around 1 percent of firm revenue today, are expected to climb significantly next year and beyond.

Next step: If your firm has deployed agents or automated workflows, give each one a clear owner and review process. Turning something on isn’t a governance plan.

5. Proof Over Promise

Enough time has passed since generative AI arrived in legal that vendors, Draftwise included, are being asked to prove impact, not just describe it.

Igor Labutov explained what that takes: “In order for us to objectively measure the impact of these tools on your data and against an outcome, it has to be done objectively against what's being achieved right now, against the output that humans have.”

Our own CEO, James Ding, put it more simply: “The point is not data science. The point is better decision-making.”

Alexis Mitchell of Syllo added a practical caution for anyone testing these tools: “You have to make a plan for what to evaluate. Evaluating on a live matter will give you information on usability or your workflow, but it isn't a good way to measure accuracy.”

Next step: Choose a repeatable task, such as first-pass contract review or diligence summaries, and establish a baseline for time and accuracy before introducing AI. Then test the same task in a controlled environment and compare the results. That side-by-side comparison tells you more about impact than login counts or usage statistics alone.

The Bigger Picture

Beyond the sessions, this year’s ILTACON felt like a good snapshot of where the industry stands. There was real energy in seeing people from across legal work through difficult problems together, and it wasn’t only coming from the big, well-funded names. Some of the newest startups on the floor were bringing genuinely creative ideas to the table.

Perhaps the clearest signal was how often conversations came back to what AI could change about the business of running a law firm, rather than which features shipped this quarter.

As Harvey CEO Winston Weinberg put it in his keynote, “This is the year of transformation, and it's about what will change in the industry. It's less about the product roadmap and more about what it will do to the profession.”

The Bottom Line

By the end of the week, the question was less about whether AI can produce useful legal work and more about what firms need around it to get consistent value.

A few priorities came up again and again:

• Data structured to connect information, not just search it
• Change management that starts with people before the tool is chosen
• Tools matched to specific workflows rather than a single firm-wide approach
• Clear ownership and oversight of what agents are doing
• Evaluation plans and outcome metrics are defined before rollout, not after

That’s where much of the work now sits.