By Raghu Ramanathan
President, Legal Professionals at Thomson Reuters
When a lawyer relies on a source to advise a client, when a legal department builds a workflow around a platform, when a compliance team puts their risk strategy on the line, they are making a bet on a partner they believe will get it right. Every single time. That trust is not given. It is earned over decades, and it can be lost in a moment.
That is the lens through which my team and I approach everything we do. We are not observers of a transformation happening around us. We are changemakers, and we hold ourselves to that standard.
The question that actually matters
Our 2026 AI in Professional Services Report makes something clear: the window for AI experimentation has closed. Enterprise-wide AI usage nearly doubled across legal, tax, and corporate sectors in a single year. More than 80% of professionals using AI today do so at least weekly. And 87% expect it to be a central part of their workflow within five years. The early adopter advantage is gone. What remains is the question of who builds on the right foundation, and who doesn’t.
Almost every AI system today can help professionals move faster. The harder question, the one that actually matters in high-stakes legal work, is whether you can stand behind the answer. Speed is the easy part. Knowing you can stand behind the answer, that is what we are focused on.
The same report found that accuracy and reliability concerns remain the top reasons professionals hesitate to trust AI. That is not a technology problem. It is a trust problem. And trust, in legal work, is not abstract – a missed citation, an unsupported analysis, an output no court would recognize are not productivity losses. They are professional failures. In some cases, they are justice failures.
Not all AI is used in the same way, and it cannot be held to a single standard. This is especially true in industries with professional liability; AI must meet a higher bar than general productivity tools. Where outputs shape legal judgments, regulatory filings, or client advice, “almost right” falls short. Thomson Reuters defines this higher bar as Fiduciary-Grade AI, the standard for how AI should perform in high-stakes professions, and the standard we build to.
What this looks like practice
CoCounsel Legal is that standard. Built on Anthropic’s Claude Agent SDK, it is a unified agentic platform that plans, retrieves authoritative content, and adapts mid-workflow — just as a senior associate would. It is built with Westlaw and Practical Law natively embedded, guided by 35 million West Key Number classifications. The lawyer stays in control throughout, able to see the agent’s reasoning, challenge its assumptions, and redirect its approach.
When Brooke Conkle, a partner at Troutman Pepper Locke, tested it on a complex TCPA question, her conclusion was that she had found not a faster tool, but a different kind of tool. That distinction is everything.
One million professionals across 107 countries have now chosen CoCounsel, including 97 of the Am Law 100 firms and 95 of the Fortune 100. That number doesn’t represent market share. It represents professionals who needed to get it right and trusted us to help them do it.
How changemakers build
Changemakers don’t wait for certainty before moving. They move when the direction is right and the foundation is sound, and then they build in partnership with customers. That is exactly how we launched the next generation of CoCounsel Legal: as a beta, with leading firms, inviting them to shape what the product becomes.
We hold ourselves to the same accountability standard we ask our customers to hold themselves to. The professionals who move the world, who uphold the rule of law, defend the wrongfully convicted, and keep the engines of commerce turning deserve AI that is worthy of the work they do.
That is what we are building. And that is why, when professionals choose who they want to build this future with, they keep choosing Thomson Reuters. That is what it means to be chosen by changemakers.
The window for AI experimentation in legal has closed. What comes next is building on the right foundation. I share what that means for legal professionals, and why the question that actually matters is not whether AI is fast, but whether you can stand behind what it produces.
