Thomson Reuters is rolling out AI agents via CoCounsel for tax, audit, and accounting needs. But, a new range of agentic capabilities for legal is also to follow ‘this year’, the tech and publishing company said.
The move follows its acquisition of Materia, an AI startup specializing in agentic systems for tax and accounting. So, it makes sense they’re starting with those areas first, then moving onto legal. That said, its main rival LexisNexis is already offering some agentic AI services for legal – and so too are a growing number of other legal tech companies.
Why is this a big deal?
Well, this is how TR explains it. They will – later this year – be bringing agentic abilities into ‘drafting, employment policy generation, deposition analysis, and compliance risk assessments’.
‘Many of these experiences already exist within CoCounsel, Westlaw, and Practical Law,’ they note, but then add that these are ‘now being upgraded with full agentic orchestration, where agents not only generate output, but plan, execute, and adapt across tools in real time.’
And as explored in other AL articles, agentic systems matter because they take us all onwards from just prompting in an ad hoc way. Agents are discrete tools that stand on their own, designed to perform a specific task, and refined by the vendors for that task – often with the ability to allow a user to further refine the output.
That in turn saves a lot of time, it also means that a lawyer doesn’t have to ‘reinvent the wheel’ every time they want to do something quite complex. They just bring up the agent (i.e. a discrete tool linked to AI systems) and apply it to the task the tool was designed to handle, (and could be aimed at a bundle of documents, a bunch of emails, legislation, case law, etc).
But, that’s not all. As TR explains, agents have some very smart aspects to them.
‘These systems are:
Built for goal-based execution across multi-step legal and compliance tasks
Designed with task-specific tool orchestration to engage both Thomson Reuters and third-party platforms
Governed by human-in-the-loop oversight for safety, accuracy, and accountability
Powered by transparent reasoning and traceable sourcing
Refined with custom LLMs trained by in-house legal, tax, and compliance experts.’
They conclude that: ‘Thomson Reuters isn’t just expanding capabilities – we’re redefining what GenAI can do in the hands of professionals.’
Last word goes to David Wong, Chief Product Officer at Thomson Reuters, who said: ‘Agentic AI isn’t a marketing buzzword. It’s a new blueprint for how complex work gets done. We’re delivering systems that don’t just assist but operate inside the workflows professionals use every day. The AI understands the goal, breaks it into steps, takes action, and knows when to escalate for human input – all with human oversight built in to ensure accountability and trust.’
Some additional thoughts here from AL:
Agentic AI is kind of a buzzword, and not every feature many legal tech companies are offering under that banner really seem to be that ‘agentic’ in nature.
This is possibly caused by a blurry definition. If agentic just means a software tool that does something, then….well….everything is an agent. So, it has to actually mean something. As noted above, it primarily has to be: custom-built for a specific task, comes to the data with a plan, can execute that plan as designed to – and that the outputs can be refined. And, if we want to get really fancy, you can orchestrate – i.e. directly manage, a bunch of agents, each with their own discrete tasks.
That this all goes back to a point AL has made a few times now about power-sharing. I.e. that if we just see, and use, AI as a little gopher via the occasional prompt, we’ll never really leverage the efficiency gains that can be achieved.
We often hear things like: ‘AI is a black box, I have no idea how or why it did X’. Well, with agents you can see their reasoning, their plan – so you can see why they did something. You can also – depending on how it’s designed, get in there and correct it so it does that thing better and/or in the way you want it to. In short, agents bring higher levels of transparency to AI – even if you still can’t see directly into how an LLM an agent is leveraging actually works for a specific question.
Any road, the field of legal AI evolves and evolves ever onwards – and still we are only just two and a half years since genAI landed in the public consciousness with ChatGPT. We have come a long way very quickly.
Exciting times. (And I don’t often say that.)
—
Legal Innovators California conference, San Francisco, June 11 + 12
If you’re interested in the cutting edge of legal AI and innovation – and where we are all heading – then come along to Legal Innovators California, in San Francisco, June 11 and 12, where speakers from the leading law firms, inhouse teams, and tech companies will be sharing their insights and experiences as to what is really happening.
We already have an incredible roster of companies to hear from. This includes: &AI, Legora, Harvey, StructureFlow, Ivo, Flatiron Law Group, PointOne, Centari, LexisNexis, eBrevia, Legatics, Knowable, Draftwise, newcode.AI, Riskaway, Aracor, SimpleClosure and more.
Cooley, Wilson Sonsini, Baker McKenzie, Gunderson, Ropes & Grey, A&O Shearman and many other leading law firms will also be taking part.
See you all there!
More information and tickets here.
P.S. there will also be a Legal Innovators New York conference this November…! See here.
Discover more from Artificial Lawyer
Subscribe to get the latest posts sent to your email.
