Harvey and LexisNexis Launch AI Agent That Drafts Motions to Dismiss End-to-End

Harvey has released a motion-to-dismiss drafting agent built jointly with LexisNexis, compressing a task that typically consumes 40 to 50 associate hours into a review-and-approve loop that resolves in hours. The workflow is the first concrete product to emerge from a strategic alliance the two companies announced in June 2025, and it signals that their partnership is moving beyond data licensing into co-engineered litigation tooling.

Harvey has spent the past two years positioning itself as the infrastructure layer for AI-native legal work, attracting large law firm clients and raising capital at a valuation that has made it one of the most closely watched companies in legal technology. The LexisNexis alliance accelerated that trajectory by giving Harvey access to primary law content and Shepard's Citations without requiring the firm to build or license a competing legal database. The earlier Ask LexisNexis integration embedded citation-backed answers directly in Harvey's interface; the motion-to-dismiss workflow is the first tool that uses those research capabilities inside a fully automated drafting pipeline. The timing matters: litigation-focused AI tools are one of the most competitive segments in the market right now, with rivals including Westlaw's AI products, CoCounsel, and a growing list of point solutions all targeting early-stage motion practice.

The workflow operates through four sequential agent stages: complaint analysis and argument identification, attorney review and strategy approval, parallel jurisdiction-filtered research across approved grounds, and final motion drafting. Research agents query LexisNexis primary law using boolean searches scoped to the motion-to-dismiss standard, supplement thin state authority with federal decisions, and surface the three to five strongest cases per argument — each carrying a full Bluebook citation, a Shepard's Signal, and a direct link to the opinion on Lexis+ with Protégé. The drafting agent adjusts automatically for forum: it files to the Rule 12(b)(6) standard in federal court and detects state-specific vehicles such as California demurrers, New York CPLR 3211 motions, and Pennsylvania preliminary objections. The output includes a Table of Contents, Table of Authorities, and citation trails linking every factual assertion to the complaint and every legal proposition to authority. Access is available to Harvey customers who have Ask LexisNexis enabled, and to Lexis+ with Protégé customers even without the integrated Ask LexisNexis feature. A motion-for-summary-judgment workflow is described as the next co-developed product in the pipeline.

Industry Implications

The release accelerates a broader shift in which legal research vendors are no longer simply selling database access but are competing to own the drafting layer of legal work. By embedding Shepard's validation inside an autonomous drafting agent, LexisNexis is defending its authority-verification moat at exactly the moment when AI-generated citations are under the highest scrutiny from courts and ethics regulators. For enterprise legal buyers and litigation departments, the workflow reframes the associate staffing calculus on early-case dispositive motions — a category of work that is billable, high-stakes, and historically labor-intensive. Competitors without a comparable research-content partnership will face structural difficulty matching citation depth and Shepard's integration without replicating deals of similar scale.

DreamLegal Perspective

Legal operations leaders evaluating litigation AI should pressure-test the MTD workflow on jurisdiction coverage depth and adverse-authority handling — two areas where automated research agents have historically underperformed human associates. The human checkpoint built into the strategy-approval stage is a meaningful liability control, but law firms should document how attorney review is conducted at each stage to satisfy evolving court-specific AI disclosure requirements. Vendors without a primary-law content partner should watch this release closely: the Harvey-LexisNexis model suggests that owning the research substrate, not just the drafting interface, is becoming the decisive competitive variable in agentic legal AI.