From skills-based tool to legal AI assistant for complex work
Thomson Reuters’ next-generation CoCounsel Legal is a legal AI assistant rebuilt to perform complex, end‑to‑end legal tasks with accuracy, security, and verifiable links to authoritative sources, setting it apart from consumer chatbots and signaling a shift toward domain‑specific enterprise legal AI for law firms. The company describes this as the most substantial overhaul since acquiring Casetext, and it now presents CoCounsel Legal less as a menu of “skills” and more as an AI colleague. Users describe a matter in plain language and the system drafts a plan, coordinates subtasks, and iterates with the lawyer instead of answering one prompt at a time. Early access starts with a toggle that lets existing customers switch between the legacy and new experiences. Thomson Reuters plans broad general availability after this early-access period, positioning the product as a central AI for law firms that want deeper integration into daily legal workflows.
Why CoCounsel Legal aims beyond consumer GPT-style tools
Thomson Reuters is drawing a clear line between CoCounsel Legal and general-purpose GPT-style systems aimed at consumers. Legal work needs answers grounded in current law, with citations that lawyers can check and trust, not responses assembled from the open internet. CoCounsel Legal ties its reasoning to Westlaw and Practical Law content maintained by large teams of attorney-editors, and every answer comes with traceable citations rather than opaque summaries. This gives law firms a legal AI assistant that is designed for motions, discovery, and research instead of casual drafting. According to Thomson Reuters, CoCounsel Legal is purpose-built so solo, boutique, and enterprise firms can “stake [their] reputation” on the results. That emphasis on accuracy, security, and verified sources highlights why many firms see generic GPT tools as useful experiments, but not as systems they can safely embed into mission‑critical legal workflows.
Enterprise legal AI, agentic workflows, and enthusiastic beta feedback
The new CoCounsel Legal reflects a broader industry turn toward agentic enterprise legal AI: systems that plan, sequence, and execute multi-step legal tasks. Thomson Reuters rebuilt the product on an agentic architecture that can coordinate research, drafting, and review within a single workspace, rather than forcing lawyers to juggle separate “skills.” This evolution mirrors advances used in Westlaw Advantage and brings the same technical teams into the CoCounsel Legal environment. In interviews, Thomson Reuters leadership said this beta was the most successful they have seen, citing strong feedback on utility, accuracy, and reliability from firms that tested the product for several weeks. Internal comments about customers having “loved” the beta experience suggest that the company sees a strong product‑market fit in enterprise legal. For law firms, that enthusiasm signals that sophisticated AI for law firms is rapidly moving from pilot projects to expected daily infrastructure.
Why Thomson Reuters is building its own LLM for legal
Alongside its Anthropic partnership, Thomson Reuters is investing in its own large language model for legal, named Thomson, underscoring a long-term commitment to domain-specific enterprise legal AI. Current CoCounsel Legal builds on Anthropic’s Claude Agent SDK, with Thomson Reuters calling its products “model‑agnostic” so they can use whichever engine performs best for a given task. At the same time, research teams from acquired labs such as Safe Sign Technologies have been developing the Thomson model, which is now in advanced testing and reported to outperform leading general models on specific legal benchmarks. The company has not yet said exactly how Thomson will be deployed, but a likely path is a hybrid stack in which some CoCounsel Legal capabilities run on Thomson and others on Claude. For law firms, that strategy promises better control over data security, legal reasoning, and long-term reliability than relying solely on third‑party consumer models.
Specialized legal AI becomes table stakes in the race for law firm adoption
The next-generation CoCounsel Legal highlights a market where specialized legal AI is becoming expected rather than optional for technology vendors that want law firm adoption. By tying AI outputs to curated legal content, providing enterprise-grade security, and supporting complex workflows, Thomson Reuters is responding to concerns raised in its own Future of Professionals research about the gap between AI hype and real value. Firms now want tools that fit their ethics rules, protect client confidentiality, and deliver measurable gains in capacity and quality. CoCounsel Legal’s early-access launch, strong beta reaction, and evolving model strategy indicate that major vendors now compete on who can offer the most reliable, verifiable AI for law firms. As more providers follow this path, general-purpose GPT interfaces are likely to play a supporting role, while specialized enterprise legal AI assistants become the primary systems that lawyers rely on for high‑stakes work.
