Legal Research AI Platforms Compared: Westlaw CoCounsel, Lexis+ AI, Harvey, and Bloomberg Law AI

Scope and Comparison Criteria

This matrix covers four platforms that have positioned AI-assisted legal research as a core product offering: Thomson Reuters' Westlaw CoCounsel, LexisNexis' Lexis+ AI, the law-firm-focused Harvey, and Bloomberg Law's Bloomberg Law AI. All four are actively marketed to practicing attorneys and have documented feature sets as of Q2 2026. General-purpose LLMs not specifically positioned for legal research are excluded.

The eight criteria below were selected because they represent the dimensions that most directly affect procurement decisions and professional responsibility compliance — not because they are the only dimensions that matter. A firm evaluating these platforms for, say, international arbitration research will need to weight corpus coverage and jurisdiction scope differently than one doing US federal regulatory work.

• Primary corpus: the underlying legal database the AI draws from for research queries
• Citation accuracy: documented hallucination rate or citation verification approach, where published
• Data retention: whether query text and uploaded documents are retained post-session, and for how long
• Confidentiality model: whether user data is used for model training
• Deployment model: cloud-hosted, on-premises, or hybrid
• Pricing structure: per-seat, usage-based, or bundled with existing subscription
• Jurisdiction coverage: whether non-US sources are substantively included
• Integration surface: how the AI layer connects to the underlying research platform and external tools

Side-by-Side Matrix

Platform Profiles

Westlaw CoCounsel

Thomson Reuters launched CoCounsel as a branded AI layer over the Westlaw database, initially through an acquisition of Casetext in 2023 and subsequent integration. By Q2 2026, CoCounsel is embedded directly in the Westlaw.com interface and available as a Microsoft Word add-in.

The primary research workflow involves submitting natural-language queries — "find federal circuit cases where a court applied the substantial similarity test to software interfaces" — and receiving a synthesized answer with inline citations linking back to verified Westlaw documents. The RAG architecture means responses are anchored to the Westlaw corpus rather than generated from a model's parametric memory, which reduces (but does not eliminate) the risk of fabricated citations.

Where CoCounsel has a clear advantage is corpus depth for US legal research. Westlaw's case law coverage, especially for older federal and state decisions, remains the broadest of any platform in this comparison. The Shepard's equivalent — KeyCite — is also natively integrated, so citation validity checking happens within the same workflow.

The pricing model is the main friction point for smaller firms. CoCounsel is an add-on to existing Westlaw subscriptions, not a standalone product, and Westlaw's base pricing is already among the highest in legal research. Firms without an existing Westlaw relationship face a significant cost barrier to entry.

Lexis+ AI

LexisNexis built its AI research layer directly into the Lexis+ platform, branded as Lexis+ AI. The product competes directly with CoCounsel on the core research workflow: natural-language queries answered with cited, corpus-grounded responses.

One differentiating feature is Shepard's integration within the AI workflow. When Lexis+ AI returns a cited case, the Shepard's signal is surfaced alongside it — negative treatment, overruled status, and citing references are visible without a separate lookup step. This is a meaningful workflow improvement for litigators doing validity checks at scale.

Lexis+ AI has also invested in document upload functionality, allowing attorneys to upload a contract or brief and ask questions against it alongside the broader Lexis corpus. This hybrid of document AI and database research is a feature CoCounsel also offers, but the implementations differ in how they handle privilege and data isolation.

Like CoCounsel, Lexis+ AI is bundled with the Lexis+ subscription rather than priced as a standalone product. The practical implication: if your firm already has a Lexis+ enterprise agreement, the AI layer is likely included or available at a negotiated increment. If you don't, the entry cost is the full Lexis+ subscription.

Harvey

Harvey occupies a different position in this comparison. It is not a legal database company that added AI — it is an AI platform built for law firms that connects to the firm's own data sources and, optionally, external legal databases. This distinction matters for how you evaluate it.

For legal research specifically, Harvey's value depends heavily on what sources it's connected to. A firm that has integrated Harvey with a Westlaw or Lexis subscription via API can run research queries against that corpus. A firm using Harvey standalone gets the model's general legal knowledge plus whatever internal documents are in scope — which is not the same as querying a verified case law database.

Harvey's strongest differentiation is on the enterprise data side: dedicated model instances, configurable data retention, and a deployment architecture designed for firms with strict data governance requirements. It has also invested in practice-area-specific fine-tuning, with configurations for M&A, litigation, and regulatory work.

The pricing model is enterprise-only, with no self-serve tier. Smaller firms will find it difficult to evaluate without committing to a sales process. That said, for large firms with complex data governance requirements and an appetite to build a custom AI stack, Harvey's architecture is more flexible than the Westlaw or Lexis offerings.

Bloomberg Law AI

Bloomberg Law's AI features are embedded in the Bloomberg Law terminal interface and web platform. The product is strongest for practitioners who need transactional and regulatory intelligence alongside case law — M&A practitioners, securities lawyers, and in-house counsel in regulated industries.

Bloomberg Law AI's corpus includes the Bloomberg Law database, which has strong coverage of SEC filings, regulatory agency guidance, transactional precedents, and news. For pure appellate research or state court coverage, it is generally considered weaker than Westlaw or Lexis.

The pricing structure is Bloomberg Law's biggest practical advantage for existing subscribers: AI features are bundled into Bloomberg Law subscription tiers rather than priced as a separate add-on. For a firm already paying for Bloomberg Law access, the incremental cost of the AI layer is lower than adding CoCounsel or Lexis+ AI to an existing subscription.

Third-party integrations are more limited than the other platforms. Bloomberg Law AI does not have the same depth of Microsoft 365 integration or API surface as Harvey or the Thomson Reuters / LexisNexis offerings.

Data Retention and Confidentiality: What the Policies Actually Say

All four platforms have published statements asserting that user query data is not used to train shared models. This is a baseline expectation for enterprise legal AI, not a differentiator. The more operationally relevant questions are: how long is query data retained, who within the vendor organization can access it, and what happens to uploaded client documents?

Harvey's architecture is the most configurable on this dimension: enterprise deployments can be structured with zero-data-retention policies and dedicated infrastructure. This matters for firms handling highly sensitive matters where even transient retention of query text raises privilege concerns.

Thomson Reuters and LexisNexis both operate under enterprise data governance frameworks that have been audited for SOC 2 compliance. Bloomberg Law's data policies are governed by Bloomberg's broader enterprise data handling standards. All three are cloud-hosted SaaS products with no on-premises deployment option.

Fit Assessment by Firm Profile

No single platform is the right choice across all firm types. The decision depends on existing subscriptions, practice area focus, data governance requirements, and budget.

Professional Responsibility Considerations

Using any of these platforms for legal research does not eliminate the attorney's obligation to verify citations independently. All four platforms use RAG architectures that ground responses in their respective corpora, which reduces hallucination risk compared to general-purpose LLMs. But "reduced risk" is not "no risk." Documented hallucination incidents involving legal AI platforms have resulted in court sanctions and bar complaints.

The confidentiality dimension — specifically, whether uploading client documents to a cloud-hosted AI platform constitutes a disclosure that triggers Model Rule 1.6 concerns — has been addressed in ethics opinions from several state bars. The general position across opinions reviewed through Q2 2026 is that use of cloud-hosted legal research tools with appropriate data processing agreements does not per se violate Rule 1.6, but attorneys must take reasonable precautions to prevent unauthorized disclosure. What counts as "reasonable" depends on the sensitivity of the matter and the vendor's documented security controls.

What This Matrix Does Not Cover

• Contract review and redlining capabilities — all four platforms have some contract AI features, but those are not evaluated here
• Drafting assistance — Harvey in particular has significant drafting functionality that is outside this comparison's scope
• eDiscovery — none of these platforms are primarily eDiscovery tools
• Benchmark test results — no independently administered, peer-reviewed accuracy benchmark for legal research AI was publicly available as of this review date
• Pricing figures — all four platforms negotiate enterprise pricing; published list prices, where they exist, are not reliable indicators of actual contract terms