If you want to play an instrument, you have to practice. If you want to get fit, you have to build a program. And if you want AI to actually work in your organization, you have to prepare for it first.
This was the consensus among legal operations leaders at CLOC 2026: successful AI initiatives require real preparation. That's the nature of change management, and there's no shortcut around it.
Stacy Lettie, VP of Global Legal Operations and Transformation at Organon, was very clear: "There is this urgency with AI. Everybody feels like they need to go out there and get a tool because that's going to make them feel better. Because they're going to feel like, well, we're doing something. We're out there. We're doing AI. We implemented this. We can go back and tell our CEO we are doing AI. But the data is showing adoption isn't as great as it needs to be."
That's the heart of the challenge. You can't announce that you're "doing AI" and expect people to follow. You have to lay the groundwork first.
Here are five things to do before you launch an AI initiative:
1. Define the problem before you buy a tool. You shouldn't be shopping for tools before you know what problems they need to solve. Alexander Shusterman, Staff Technical Program Manager, CLO AI, and Carolyn Wakulchik, Manager, CLO Operations— both from Uber— offered this: "Scope the risk area. What do you care about the most? If you ask everybody, they're going to tell you they want the sun, moon, and stars in the solution, and that is a surefire way to make sure that it doesn't work correctly. Better to start small and focus."
2. Get your data in order. AI is only as good as the data that fuels it. Stacy Lettie described the problem: "You have not gotten your data together. You have not cleaned your data. You have not even put all of your documents in one SharePoint folder, for gosh sakes. And you all of a sudden go, 'Oh wait, I can't do this, and now I've got to stop, and it's going to take me eighteen months to clean my data.'" Don't let this happen to you. Before anything else, make sure the data your AI will use is organized and accessible.
3. Build a governance framework. Guardrails aren't optional. They’re what make AI adoptable for the rest of the organization, and keep your company and your customers safe. Think through policies for accuracy, data protection, and appropriate use. Governance has to be implemented cross-functionally, so involve the right stakeholders early, and make it easy for team members to follow the policies you put in place.
4. Plan for change management. AI initiatives stall when no one owns adoption. Lawyers are risk-averse by training, and they won't use tools they don't understand or trust. Alexander Shusterman from Uber was very frank: "Just like any other legal operations project, the hardest part here is not the tech and it's not the prompting. It is actually the change management piece."
5. Agree on how you'll measure success. Before the initiative launches, define what success looks like, and get alignment from the executive team on those metrics. This discipline helps you scope the problem well and makes it much easier to demonstrate value in the future.
Successful AI initiatives don't happen by accident; they're built on preparation. These steps have been field-tested by legal operations leaders who've done the work. If you’re looking for a framework to start planning your AI projects, this is a good place to start.
.jpg)
