Next Gen CoCounsel To Offer ‘Fiduciary-Grade’ Legal AI

Thomson Reuters’ Ragunath Ramanathan, President – Legal Professionals, has announced that the ‘next generation’ of CoCounsel Legal, now available in Beta, will provide ‘fiduciary-grade AI’ and work just as well as a senior associate.

‘The next generation of CoCounsel Legal is a unified agentic platform that plans, selects tools, retrieves authoritative content, and adapts mid-workflow just as a senior associate would, not a first-year waiting for the next instruction,’ he said.

Much of this is built using Anthropic’s Claude Agent SDK. To get hold of these new capabilities you have to join the Beta program, which has a waitlist. The announcement slipped out in a blog post by the Thomson Reuters executive this week.

They said that by building together via the Beta program ‘we’re inviting customers to help shape what CoCounsel Legal becomes’.

(And of course, readers will be well aware that at the same time Anthropic is moving directly into the legal sector now – although what it doesn’t have is TR’s huge curated legal data collection.)

Sounds intriguing. So, what is on offer here? The short answer is that it’s all about trust and accuracy – a topic that the recent Sullivan & Cromwell debacle has brought into sharp focus.

Ramanathan states that as well as improving the reasoning capabilities of CoCounsel when it comes to the AI exploring a research task, that they have ‘gone further to protect the integrity of that reasoning, with patent-pending tools for citation integrity and output verification’.

These improvements are centred on two key areas, he explained:

• ‘Verification and grounding as system primitives. Authoritative retrieval, explicit source handling, and verifiable citation flows are product infrastructure – not post-processing or marketing language.
• Patent-pending citation integrity. Our citation ledger architecture creates a session-verifiable evidence trail for legal citations, ensuring the agent can only cite what it actually retrieved and enabling trustworthy, clickable references in output.’

(Note: it’s not 100% clear if these two developments are already available to regular users of CoCounsel Legal, or if it’s just for the new Beta program customers. AL has asked TR for more info.)

And: ‘Our leading evaluation framework encodes quality at each step. This means before any capability ships; we measure it. Licensed attorneys, including our Practical Law editors, define what the correct output looks like for each task type.

‘Every new capability must demonstrate measurable improvement against that benchmark before it reaches production. The framework evaluates not just final outputs, but the full chain of reasoning that produced them, because an agent that arrives at the right answer through flawed reasoning cannot be trusted to do so consistently.’

He continues: ‘This is fiduciary-grade AI; outputs grounded in authoritative content and customer context – making verification part of the system’s architecture rather than an afterthought.’

‘In a profession where a single missed citation can cost a client their case, defensibility isn’t a nice-to-have. It’s the whole point. In a profession where a single missed citation can cost a client their case, defensibility isn’t a nice-to-have. It’s the whole point,’ he concluded.

—

Conference Advertisement:

A Legal Tech Conference For All of Europe – Legal Innovators Europe – Paris – June 24 and 25.

Express route to your ticket here.

And, 

Legal Innovators California, the landmark West Coast legal tech event, will take place on June 10 and 11, in the heart of the Bay Area, the home to many of the world’s leading AI businesses – and plenty of legal tech pioneers as well! More information and tickets here.

Express route to your Legal Innovators California June 10th and 11th ticket here.

—

(article continues)

Is this a big deal?

Well, it’s interesting that legal AI accuracy has zoomed right back into legal sector consciousness after a long period where it was seen as sort of ‘done now’. Clearly accuracy is not ‘done’, there is more work needed and Thomson Reuters is focusing on that.

Also of note is the fact that Sullivan & Cromwell – so far, at least as understood – has not publicly named the AI tool(s) that created the hallucination that got noticed. Let’s hope it was not one of the more famous legal AI / legal research companies..!

Regardless, all of this will help to push accuracy up the legal tech agenda again and the latest moves by Thomson Reuters therefore come at a good time.

—

P.S. to join the Beta waitlist, get in touch with TR. They added that if you join you can expect:

• ‘A simpler, conversation‑driven experience with no tool‑switching or prompt engineering. Just describe what you need and CoCounsel Legal figures out how to get it done.
• Powerful, end‑to‑end execution from a single plain‑language request—research, strategic guidance, analysis, and drafting all delivered in minutes from a single query.
• A chance to shape the future of legal AI, share direct feedback, and help define the next generation of CoCounsel Legal.’

—

Link to the original Blog by Ragunath Ramanathan here.

—