The system routes prompts across models, agents and legal sources, then produces Word, Excel and PowerPoint files for lawyer review.
By Ryan Merket · Published
Primary source: LexisNexis
Why it matters #
LexisNexis and Thomson Reuters are racing to own the workspace where AI-generated legal work is grounded, checked and converted into client-ready files.
LexisNexis unveiled new agentic capabilities for Lexis+ with Protege on August 24th, extending its legal AI platform across research, analysis, drafting and review while preserving the context of a matter between each step.
The release came four days after Thomson Reuters launched the next generation of CoCounsel Legal, built around an almost identical strategic premise: lawyers should be able to move from a legal question to cited, reviewable work without manually carrying information among separate research, drafting and document-analysis tools.
That timing puts the two largest legal information incumbents in direct competition over the software layer where lawyers finish their work. Both are using proprietary legal databases, citation systems and customer documents to make general-purpose AI models useful inside matters where unsupported claims can create professional and financial liability.
Sean Fitzpatrick, CEO of LexisNexis Global Legal, has spent more than two decades inside the company after joining in 2005 from McKinsey. He led the commercialization of Lexis+ AI and the introduction of Protege, giving him the job of converting LexisNexis's research franchise into a workflow product before newer legal AI vendors separate lawyers from the company's underlying content.
"Legal professionals should not have to spend their time managing AI, switching among tools or repeatedly explaining how they want work performed," Fitzpatrick said in the announcement.
One prompt, several systems behind it
In the expanded Lexis+ with Protege, a lawyer describes the intended result instead of selecting an individual AI tool or workflow. LexisNexis says the platform then chooses among supported models, agents, specialized skills and approved sources while carrying the matter's context into subsequent tasks.
A single instruction can initiate a multi-step drafting process and produce formatted Word documents, Excel spreadsheets or PowerPoint presentations. The system supports document-aware editing, structured tables, track changes and suggested follow-up actions. Lawyers can also select organizational files from a connected document management system and use them alongside LexisNexis materials.
The grounding hierarchy is central to the product. LexisNexis legal sources remain the default, while open-web and customer sources are used when selected. Citations link back to authority and can be checked with Shepard's citation intelligence. LexisNexis calls the resulting files "review-ready," language that keeps the lawyer responsible for checking and approving the work.
Users can allow the platform to select what LexisNexis considers the best model for a task or choose a preferred supported model. The company also plans to preview its own models later in 2026. Customer-created skills and workflows are on the roadmap; the current platform can use skills supplied by LexisNexis and its technology partners.
The August release extends a product overhaul already underway. Lexis+ AI was renamed Lexis+ with Protege in February 2026, and LexisNexis added structured agentic workflows, drafting and citation verification in subsequent releases. In May, the company introduced Protege Work, which converts a stated legal goal into a visible execution plan and routes it to the appropriate skill or workflow.
RELX, the parent company of LexisNexis, said in its 2025 annual report that Protege followed more than $1 billion in technology spending over a decade. LexisNexis began experimenting with agentic AI in 2023 and introduced Protege in January 2025, according to the report.
Legal AI vendors converge on finished work
The competitive field has moved quickly from chat interfaces toward systems that plan and execute longer assignments. Thomson Reuters says its new CoCounsel Legal can move from research and issue analysis to a first-draft brief using Westlaw, Practical Law and KeyCite. It also preserves matter context in workspaces and drafts inside Microsoft Word.
Legal AI startup Harvey made a similar move in May, introducing more than 500 task-specific agents that it says can complete longer processes and produce Word, Excel and PowerPoint files. Harvey also lets law firms build agents around their own practices and institutional knowledge.
LexisNexis is responding with the assets that startups would have difficulty recreating: its legal corpus, Shepard's citation system, existing customer relationships and access to internal firm documents. The company says its systems draw on more than 200 billion legal and business documents and records. That figure and the claim that more than 4,000 technologists and legal specialists develop the platform are company-supplied.
The product battle now centers on orchestration and control. Model providers can supply stronger reasoning, while legal publishers decide which sources, permissions, citation checks and workflows surround those models. By hiding model selection and agent routing behind one prompt, LexisNexis is betting that lawyers will value a managed legal workspace over assembling their own collection of AI tools.