We respect attorney-client confidentiality. No tracking pixels in our emails.
LexisNexis Protégé launched with bold claims about brief answering and memo drafting. After months of real-world use, here's what attorneys actually think about accuracy, pricing, and how it stacks up against CoCounsel and Vincent AI.
A partner at a mid-size litigation firm in Chicago ran a direct test in early 2026: she gave LexisNexis Protégé and CoCounsel the same research question — a circuit split on personal jurisdiction after Mallory v. Norfolk Southern — and compared the outputs. Protégé returned a well-structured memo with accurate citations to the relevant circuit decisions and a clear summary of the split. CoCounsel returned a similarly structured memo, but with two additional cases the partner hadn't encountered that proved directly on point. The test wasn't scientific, but it framed the question she — and thousands of other attorneys evaluating Protégé — are now asking: is Protégé's deep Lexis integration enough to offset the gap in AI model quality?
This review covers what Protégé actually does, how it differs from the existing Lexis+ AI product, pricing structures, accuracy observations across research tasks, and a direct comparison with CoCounsel and Vincent AI.
LexisNexis has operated in the legal research market for decades, and its content library — case law, statutes, regulations, secondary sources, and news — is one of the most comprehensive in the industry. The company's challenge entering the AI research assistant era was not content; it was model quality and interface design. The original Lexis+ AI, launched in 2023, was widely perceived as a first-generation product: useful for simple queries but limited on complex multi-step research and prone to the citation errors that AI hallucination risks create.
Protégé was announced in late 2025 as a substantively different product — built on a more capable large language model architecture with deeper integration into the Lexis content layer. The name signals the positioning: a junior associate-level AI that can handle research assignments, draft memos, and answer brief questions with the same authoritative sourcing that has characterized Lexis content for decades.
The timing is significant. Protégé launched into a market where CoCounsel (formerly Casetext's CoCounsel, now operating independently after the Thomson Reuters acquisition disruption) had established itself as the benchmark for AI legal research quality. Vincent AI, developed by a team with deep litigation expertise, had carved out a strong position for litigation-specific research and deposition prep. Westlaw Precision had the same content-depth advantage that Protégé claims, but through a different AI architecture. Protégé needs to differentiate on all three dimensions simultaneously.
The attorney community's early response has been cautious but interested. Firms with existing Lexis enterprise agreements are evaluating whether Protégé upgrades warrant contract renegotiation. Solo and small-firm practitioners are asking whether the pricing delta over Lexis+ AI is justified. Large firms with dedicated research librarians are testing whether Protégé can reduce the volume of routine research requests that consume librarian time.
Protégé handles four primary task types: brief answering (direct questions about legal standards, elements, and requirements with sourced answers), research memo drafting (multi-page structured memos synthesizing case law across jurisdictions), document analysis (reviewing uploaded briefs or contracts and answering questions about their content), and citation validation (checking whether cited cases remain good law and whether they support the proposition for which they're cited).
The difference from Lexis+ AI is most visible in multi-step research tasks. Lexis+ AI handles single-query searches well but struggles when a research question requires synthesizing information across multiple source types — case law plus secondary sources plus regulatory guidance, for example. Protégé handles this more coherently, maintaining context across a research session and producing outputs that read more like attorney-drafted memos than search result summaries.
Document upload and analysis is new to Protégé and represents a meaningful capability expansion. Attorneys can upload a brief from opposing counsel and ask Protégé to identify every cited case, verify each citation's current validity, and flag any cases where the cited proposition appears to be overstated or where more recent authority cuts against the cited position. This is the kind of task that previously required several hours of attorney time.
Protégé's citation accuracy on established federal circuit and district court case law is strong — this is where the Lexis content integration pays off most clearly. For research questions anchored in well-developed bodies of doctrine (Fourth Amendment suppression standards, summary judgment procedure, contract interpretation canons), citation accuracy runs high and the cases returned are generally the right ones.
The accuracy picture changes on two categories of queries. First, recent regulatory changes and agency guidance from the past 12-18 months are underrepresented or occasionally misstated — the training data lag that affects all AI legal tools is not fully offset by Protégé's real-time content access, because the AI's synthesis layer doesn't always correctly incorporate very recent materials. Second, state-specific statutory research shows more variance than federal case law research, particularly in smaller jurisdictions where the Lexis corpus is thinner.
Attorneys should treat Protégé outputs as a strong first draft requiring verification, not as an authoritative final product. This is true of all current AI legal research tools, but it bears specific mention here because Protégé's confident presentation style can create false assurance.
CoCounsel's core strength is AI model quality — the underlying model is better at synthesizing complex, multi-jurisdiction research questions and at recognizing when a research question has no clean answer and requires hedged analysis. In direct comparisons of research memo quality on complex federal civil procedure questions, experienced litigators consistently rate CoCounsel outputs as more analytically sophisticated.
Protégé's advantage is content depth and integration. Shepard's integration is native and seamless; CoCounsel's citator integration is functional but requires more deliberate use. For secondary sources — law review articles, treatises, practice guides — Protégé's access is more comprehensive. For attorneys whose research workflows are built around secondary source navigation, this matters.
Pricing: CoCounsel prices as a standalone subscription starting around $100/month per seat for basic tiers. Protégé is available as an add-on to existing Lexis subscriptions, with pricing that varies significantly based on existing contract terms. For firms already paying for enterprise Lexis access, Protégé may represent a modest incremental cost. For firms without Lexis, the total cost of entry is substantially higher than CoCounsel.
See the CoCounsel vs Casetext comparison for more detail on the competitive dynamics in this market segment.
Vincent AI is purpose-built for litigation practice and particularly strong on deposition preparation, witness research, and damages analysis. It is not a general-purpose legal research tool. Protégé is broader in scope but shallower in litigation-specific functionality.
For a litigation-focused firm, Vincent AI and Protégé are complementary rather than directly competing. Vincent AI handles the litigation workflow; Protégé handles general research. The question is whether a firm wants to pay for both, or whether one platform can adequately cover both use cases. Currently, neither does the other's core function as well as the specialist tool.
Protégé pricing tiers integrate with Lexis subscription levels. Solo practitioners on basic Lexis plans pay a modest monthly add-on. Mid-size firm enterprise agreements bundle Protégé at a negotiated rate that is generally more favorable than the list price for standalone access. The exact pricing varies by contract, and LexisNexis's sales process for enterprise customers involves significant negotiation.
One important note: some features that are highlighted in Protégé marketing materials — including certain document upload capacities and the most advanced memo drafting capabilities — are only available at higher subscription tiers. Attorneys evaluating Protégé should clarify which tier includes the specific features that drove their interest before signing.
Scenario: Circuit split research for a federal appellate brief
An appellate litigator needs a memo on the current state of the circuit split regarding the "plausibility" pleading standard's application to fraud claims under Rule 9(b).
Using Protégé: The attorney opens a new research session, asks the brief question directly, and receives an initial answer with citations. She then asks follow-up questions to drill into specific circuits, requests a formatted memo, and asks Protégé to verify all citations via Shepard's. Total time: approximately 25 minutes for a research memo she would previously have spent 3-4 hours drafting.
The output requires review — two cases in the memo require her to verify that the cited proposition accurately reflects the holding, not just the discussion. Both check out. One circuit's position is characterized as "unsettled" when in fact there is a clear controlling case — she catches this and corrects it.
Assessment: Protégé saves substantial time on this task. The errors require experienced attorney review to catch. A junior associate using Protégé without supervision could miss them. The tool is most valuable as an accelerant for experienced attorneys, not as a replacement for legal judgment.
LexisNexis Protégé — Best for attorneys already deeply embedded in the Lexis ecosystem. The content integration advantage is real and most valuable for research workflows that rely heavily on secondary sources and Shepard's.
CoCounsel — The benchmark for AI model quality in legal research. Better choice for complex multi-jurisdiction research and for firms not already committed to Lexis infrastructure.
Vincent AI — Superior for litigation-specific tasks: deposition prep, witness research, damages analysis. Complements rather than competes with general research tools.
Westlaw Precision — Protégé's closest structural analog in the Thomson Reuters ecosystem. Firms with Westlaw commitments should compare directly before any platform switch.
Fastcase — Worth considering for solo and small-firm attorneys who need strong case law access with AI features at a lower price point than the major platforms.
Q: If my firm already pays for a Lexis enterprise subscription, is upgrading to Protégé worth it?
A: Likely yes, if your team uses Lexis for active research rather than just as a citation checker. The memo drafting and document analysis features alone justify the upgrade cost for attorneys doing more than 10 hours of research monthly. Get a pricing quote tied to your specific contract before deciding.
Q: How does Protégé handle prompt engineering — do attorneys need to learn specific query structures?
A: Less so than earlier AI research tools. Protégé handles natural language research questions well. However, attorneys who learn to specify jurisdiction, time period, and desired output format in their initial query get significantly better first-draft outputs.
Q: Can Protégé be used for transactional research, or is it primarily a litigation tool?
A: It handles transactional research — contract interpretation precedents, regulatory compliance questions, M&A-related legal standards — but it's less differentiated there than in case law research. Transactional teams may find Harvey AI or CoCounsel's transactional modules more purpose-fit.
Q: How should firms handle the hallucination risk when supervising junior attorneys using Protégé?
A: Establish a verification protocol: every case cited in a Protégé output must be manually checked for accuracy of the cited proposition before the output is included in any filing or client deliverable. Build this into your AI use policy. The AI hallucination risk is manageable with process controls but not ignorable.
Q: Does Protégé support collaborative research across a team, or is it single-user?
A: Enterprise tiers support shared research sessions and output sharing within a matter context. The collaboration features are less developed than some purpose-built matter management tools, but functional for basic team research workflows.
This article reflects independent editorial analysis. LawyerAI does not accept payment for editorial coverage. Tool scores are based on methodology described in Our 5-Dimension Methodology. Last reviewed: 2026-06-23.
