What the Next-Gen CoCounsel Legal AI Actually Is
Thomson Reuters’ next-generation CoCounsel Legal AI is an enterprise legal software assistant that combines a proprietary legal LLM, DeepJudge integration, and curated research content to deliver accurate, verifiable answers for complex professional work. Rather than acting as a general chatbot, CoCounsel Legal is designed as an “agentic” system that runs multi-step tasks such as document review, legal research, and contract analysis against trusted, governed data sources. The new release is the most substantial reworking of CoCounsel since Thomson Reuters acquired Casetext, and it now emphasizes strict alignment with authoritative content from Westlaw and Practical Law. By replacing a pure reliance on external models with its own Thomson LLM, alongside model-agnostic support for tools like Claude, Thomson Reuters is repositioning CoCounsel as fiduciary-grade infrastructure for law firms that need accuracy, security, and audit trails rather than generic AI answers.
DeepJudge Integration: Turning Firm Knowledge into Searchable Intelligence
The newly available DeepJudge integration is central to CoCounsel Legal’s appeal for law firms that want AI grounded in their own work product. DeepJudge connects to a firm’s document management systems and knowledge repositories, applying semantic search so CoCounsel can surface precedent, briefs, and internal templates relevant to a specific fact pattern. Importantly, this institutional knowledge stays governed where it already lives; there is no separate knowledge store to build or migrate. Existing permissions and ethical walls are preserved, which means access rights remain intact while AI results draw from both internal documents and Thomson Reuters’ content. For the first time, lawyers can see prior firm work, Westlaw legal research, and Practical Law market standards in one CoCounsel Legal experience, giving what DeepJudge’s team describes as a 360° view of a matter. This DeepJudge integration pushes legal tech AI tools toward being live knowledge systems rather than static research portals.
Why a Proprietary Legal LLM Matters for Enterprise Law
Thomson Reuters’ decision to build its own proprietary legal LLM, called Thomson, signals a strategic pivot toward dedicated AI infrastructure for regulated professionals. The company has tested Thomson against leading general models and reports that it already outperforms them on specific legal tasks. At the same time, CoCounsel remains model-agnostic, with agentic workflows that can run on Thomson, Anthropic’s Claude, or a mix of both. This hybrid architecture gives Thomson Reuters flexibility to optimize for performance, cost, and risk while keeping tight control over how legal data flows through the system. For law firms, the key benefit is confidence: the same technologists and researchers behind Westlaw Advantage now train and tune the LLM on curated, citation-rich legal content. That alignment between content and model is difficult for consumer AI tools like GPT, which rely on public internet data and lack domain-specific governance.
Beyond GPT: Complex Legal Tasks CoCounsel Can Handle
CoCounsel Legal is designed to tackle tasks that general-purpose models struggle with, especially when accuracy and verification are non-negotiable. Litigation teams can use it for discovery preparation, summarizing large volumes of documents, and generating draft motions tied to specific authorities. Solo and boutique law firms use CoCounsel Legal to perform research grounded in Westlaw and Practical Law content maintained by more than 1,500 expert attorney-editors, instead of scraping the public web. Every answer arrives with citations lawyers can check, supporting fiduciary-grade use in court filings and client advice. Consumer tools like GPT can draft emails or summarize news, but they cannot reliably account for jurisdiction, current law, or firm-specific precedent. CoCounsel Legal fills that gap by combining structured legal LLM reasoning, retrieval from trusted databases, and DeepJudge-powered searches of internal work product, turning AI into a force multiplier instead of a risky shortcut.
What This Means for the Future of Legal Tech AI Tools
The CoCounsel–DeepJudge integration highlights a broader shift in legal tech AI tools from generic chatbots toward specialized enterprise platforms designed for regulated services. Thomson Reuters’ “fiduciary-grade AI” framing underscores that law firms want more than productivity gains; they need security, auditability, and clear responsibility for outcomes. By tying CoCounsel Legal to a proprietary legal LLM and deeply integrated knowledge systems, Thomson Reuters is setting a template for how enterprise legal software can blend public models, private models, and controlled content. According to Thomson Reuters CEO Steve Hasker, the early beta of next-gen CoCounsel has been “the most successful beta trial” he has seen, with strong feedback on accuracy and reliability. If that reception holds, law firms may begin to view AI not as a consumer add-on but as core infrastructure, comparable to research platforms and document management systems.
