LexisNexis Protégé Is Now Live In EvenUp, Letting PI Lawyers Query Case Law Using Facts From Their Case Records

The useful part of Protégé in EvenUp is the fact-selection step, and nobody outside the two vendors has tested it

LexisNexis and EvenUp, an AI platform for personal injury lawyers, announced an alliance in September. The Protégé integration is now live, having shipped earlier this month, according to LawNext's Bob Ambrogi, who watched a demo last week. EvenUp's Companion assistant can call Protégé over the Model Context Protocol. It sends the legal question plus case facts pulled from EvenUp's extracted data, and it returns an answer grounded in Lexis case law and statutes. Both subscriptions are required. The reverse direction, where a lawyer in Protégé pulls facts out of EvenUp, is promised but has no date. The demo was led by EvenUp's Saam Mashhad and LexisNexis's Liz Christman, using a sample Florida collision case. It was not an independent test.

What does the integration actually do?

A lawyer clicks a Protégé icon in Companion's prompt box and asks a legal question. Companion packages the relevant case facts with it, and Protégé answers inside the chat with linked citations.

Each statement carries a citation that expands to the supporting excerpt. Cited authorities appear with Shepard's signals and open in Lexis+ for deeper checking. A lawyer can carry the conversation into Lexis+ with Protégé, with the EvenUp context attached.

Can lawyers trust the facts Companion chooses to send?

Not on the evidence so far. Companion decides which extracted facts are relevant, and the coverage gives no way for a lawyer to inspect that selection or measure how often it is right.

The citation trail checks the law, not the fact set. If Companion omits a bad fact or misreads a medical record, Protégé's analysis will be well cited and still wrong for the case. Christman said lawyers' prompts are growing as they try to share more case context. That is a reasonable argument for automating the step. It is also her anonymized usage data, and no figure was offered. Whoever builds the fact package now shapes the legal answer, and the lawyer needs to see that package.

What happens to case data when it crosses to another vendor?

Targeted case facts leave EvenUp and go to LexisNexis with each query. The sources provided do not describe retention, training or logging terms for that hop.

Personal injury files hold medical and other sensitive records. A "dance" between two vendors, as Mashhad put it, means two sets of data terms. Our earlier coverage of sovereign, smaller-footprint AI for firms and of AI risk outpacing insurers' pricing applies here. Chained agent calls add parties whose handling of privileged material a firm has to understand. A selective fact package is arguably better than a bulk upload. It is still a transfer.

What should buyers ask their own vendor?

Ask whether your platform can pass structured case facts to a research tool, show you exactly what it sent, and keep a record of it. Then ask who retains that data.

Also ask for error rates on fact selection, measured on your own matters rather than a sample. The Florida demo applied a damages threshold to one plaintiff's injuries, which is a tidy case. In discovery-heavy litigation, relevance is contested, and an automated filter that picks "what matters" needs defensibility of its own. Firms that already have a cited, auditable record should expect any integration to preserve that auditability end to end.

Frequently asked questions

Do I need both products?

Yes. Use of the integration requires subscriptions to both EvenUp and LexisNexis Protégé.

Can I start in Protégé and pull case facts from EvenUp?

Not yet. The companies say that direction is coming, but the report gives no date.

Does the integration replace checking authorities?

No. It surfaces Shepard's signals and links to full analysis in Lexis+, which is where verification still happens.

Sources: LawNext, reporting by Bob Ambrogi.