Inside Thomson Reuters’ Bid to Build a Unified Legal AI Platform
The legal tech giant is consolidating its AI capabilities across platforms as the industry shifts toward comprehensive, integrated solutions.
Thomson Reuters beta launched Deep Research, marking a significant step in the company’s strategy to embed advanced AI throughout its legal research ecosystem.
The platform now integrates Deep Research, an agentic AI research capability, directly into Westlaw Advantage and Practical Law, alongside guided workflows designed to handle multi-step legal tasks. The company has also expanded CoCounsel’s multilingual capabilities, with support for French, German, Japanese, and other languages, signaling its broader global ambitions.
AI that reasons through legal questions
Deep Research enables lawyers to delegate full research questions to an AI system that plans its approach, explains its reasoning process, sources its answers, and delivers structured reports grounded in Westlaw and Practical Law content. Unlike traditional AI tools that simply respond to prompts, the system uses guided workflows that combine practical legal expertise with AI capabilities to walk users through complex processes such as drafting complaints, creating employee policies, and conducting jurisdictional surveys.
Colleen Nihill, Chief AI and Knowledge Management Officer at Am Law top 10 firm Morgan Lewis, noted that Deep Research stands out for its ability to reason through legal questions rather than simply return search results, with transparency essential for maintaining the oversight and trust lawyers need to confidently adopt AI in practice.
Embedded in existing workflows
The integration addresses a critical challenge in legal AI adoption: making advanced capabilities accessible within existing workflows rather than requiring lawyers to switch between multiple platforms. During a media briefing, Thomson Reuters demonstrated how the system can locate documents across multiple repositories, including document management systems and cloud storage, then use those documents to generate complex legal filings. The workflow provides links to relevant Practical Law resources and produces first drafts for attorney review.
Built on Westlaw’s research toolset
Mike Dahn, head of product management for Westlaw, explained that the system emulates best practices of experienced researchers by using Westlaw’s full research toolset, including Key Numbers, KeyCite, Precision Research classifications, and statutes annotations. Rather than layering generic AI on top of legal content, Thomson Reuters trained AI agents to use Westlaw’s exclusive research tools with curated, up-to-date content to move through complex legal research workflows.
Major investment in AI development
The platform is already used by over 20,000 law firms and corporate legal departments, including the majority of Am Law 100 firms and top U.S. courts. Thomson Reuters tested the Deep Research functionality with more than 1,200 customers and attorneys before the launch. The company is investing more than $200 million annually in AI development, underscoring how seriously legacy research providers are betting on generative tools.
The shift toward one-stop-shop AI solutions
The Thomson Reuters approach reflects a broader industry trend toward integrated AI solutions. The move follows similar ecosystem plays by LexisNexis and vLex, which are also racing to embed generative AI into their research environments. Rather than offering standalone AI tools that require lawyers to juggle multiple applications, legal technology providers are increasingly building comprehensive platforms that combine research, drafting, workflow automation, and document analysis within unified environments. This one-stop-shop model aims to reduce friction in legal workflows by embedding AI capabilities directly where lawyers already work, whether in research databases, practice management systems, or document editors.
The shift toward integration also addresses practical concerns about data security, billing efficiency, and training costs. By consolidating AI capabilities within familiar platforms backed by trusted legal content providers, firms can more easily manage access controls, track usage, and ensure quality standards across their practice groups.
My Take
As expected, legal tech vendors are racing to build all-in-one AI platforms that promise to handle every research, drafting, and workflow need in one seamless space. For most firms, that kind of integration will be a welcome relief from juggling tools that never quite fit together. But I keep coming back to a question that nags at the edges of this progress: if every firm uses the same AI, will their work start to sound the same?
Uniform platforms can make practice more efficient, but they also risk flattening it. When the same systems are doing the research, writing, and even suggesting strategy, the distinctive voice of each firm could fade. The differentiator may shift from human judgment to prompt engineering—and even that advantage disappears if everyone’s drawing from the same built-in prompt libraries.
It’s possible that the real edge will belong to firms that treat these universal tools as a starting point rather than the finish line. Those that train private models on their own documents, or layer custom reasoning steps over generic AI, might preserve what makes their practice unique. Everyone else could find themselves producing the same competent, predictable, and uninspired work product.
And maybe that’s the inevitable trade-off. Most of these platforms still run on the same few base models—ChatGPT, Anthropic, and a handful of others—so even the most elaborate integrations are still built on a shared brain. If that’s the case, the future of legal practice might depend less on who has access to the best AI and more on who can still think independently while using it.
