
Getty Images
Thomson Reuters intros agentic AI Deep Research system
The AI platform uses the company's legal content and tools, such as Westlaw Advantage.
Thomson Reuters on Tuesday launched a custom agentic AI system aimed at lawyers.
CoCounsel Legal with Deep Research combines legal research, essential workflow automation, intelligent document management and AI-powered legal assistance. The content and technology vendor said the Deep Research technology doesn't sit next to the work; it is embedded in it and designed to provide support in litigation, transactional work and regulatory analysis.
CoCounsel Legal with Deep Research is based on Thomson Reuters' content and tools, including the Westlaw Advantage legal research platform. Deep Research is built to reason, plan and deliver legal research results. The agentic tool also keeps human reviewers in the loop by explaining its process, sourcing its answers and building the foundation of its argument, according to the company.
Meanwhile, another tool, agentic guided workflows, applies agentic AI to high-friction legal work such as drafting privacy policies, complaints, discovery responses and employee policies.
Agentic AI and law
The new research tool comes as agentic AI capabilities have become popular in the AI market.
It is also an example of how AI technology is changing many industries.
"Like every industry, the legal profession is going to have to face the fact that AI can often and increasingly do our jobs better than we can," said James Michael Cooper, a California Western School of Law professor. "This new set of innovations is helping to drive that conversation in the agentic AI and law space."
While the legal profession is ripe for the use of AI technology, using AI closely with law also carries challenges.
Challenges with AI tools and law
One of the problems is that some lawyers who have used AI tools in research and writing of their briefs have ended up getting in trouble because of hallucinations.
For example, this week, a lawyer in Springfield, Ill. was fined after admitting that a court found that some of the cases he cited in a brief did not exist.
Because more lawyers are facing scrutiny for their use of AI tools, especially when the AI tool cites cases incorrectly, many firms will likely be attracted to a tool like Deep Research from Thomson Reuters, said Futurum Group analyst Bradley Shimmin.
"Companies will go with Thomson Reuters in lieu of [other AI tools]," Shimmin said. He added that the vendor has experience and expertise that it can apply to make the system a success if it is resilient and adaptable to its environment.
However, providing a technology-based legal expert like Deep Research can also be risky for Thomson Reuters, Shimmin continued.
"CoCounsel drafting 'stuff' definitely requires them to build a robust system," he said. "A system that can anticipate and remediate challenges that a company might not be able to anticipate or remediate on their own."
Thomson Reuters said it ensures that the information generated by Deep Research is verified.
"We make sure that our agentic process uses our research tools in Westlaw and our curated content set," said Mike Dahn, head of Westlaw product at Thomson Reuters. He added that the content set is kept up to date and analyzed by lawyer editors.
"It's not remotely like searching the open web," Dahn said. "Our agents are searching in that curated, annotated, corrected collection of law, and they're using the tools of Westlaw … to find the law very quickly."
Moreover, the Deep Research capability works similarly to how a lawyer would at the beginning of his or her research. Most lawyers learn from the documents they read and use that information to guide them to the statutes they need to look up, the company said.
Despite the precautions Thomson Reuters has taken with Deep Research, Cooper, of California Western, said it is still important for lawyers to be their own lawyers and stay involved.
"We, lawyers, still have an important role," he said. "We still have to read and reshape the output. But increasingly, we will rely on automation to drive business decisions, legal opinions and client advice."
Esther Shittu is an Informa TechTarget news writer and podcast host covering AI software and systems.