TechPrufer
Get started
Harvey
Harvey logoHarvey · Feature trail

AI Demand Letter Generation

16 posts · Aug 2024 – Sep 2026

What this trail covers

Harvey's AI Demand Letter Generation capability has been enhanced with specialized modes (Assist and Draft) tailored to different use-cases, iterative functionality for refining and expanding on initial responses, and faster processing, higher-quality outputs, and enhanced document processing capabilities. Assist Mode is optimized for summaries, analyses, and searches, while Draft Mode is purpose-built for generating and revising detailed long-form content. Iteration is supported through follow-up questions and revisions, with improved memory for long prompts and large documents. Performance improvements include a 60% reduction in hallucinations, a 23% improvement in cited source accuracy, an 80% improvement in time to first word, and better support for Excel files and multiple languages.

All Harvey trails
  1. How to Draft Discovery Requests and Responses You Can TrustThis post details the technical process of grounding and verifying AI-generated discovery requests (interrogatories, requests for production, requests for admission) to ensure they are warranted, proportional, and not interposed for an improper purpose, as required by Rule 26(g). It explains how AI tools like Harvey can assist in drafting these requests by analyzing case records, governing law, and local rules, but emphasizes that the lawyer's reasonable inquiry and verification remain critical. The post highlights the importance of tying each request to specific claim or defense elements, grounding scope in the matter's record, and ensuring proportionality by considering case needs and resources. It also discusses how AI can help red-team these requests by identifying overbroad or compound questions before they are served.
  2. How to Draft and Review an Engagement Letter With AIThis post details the technical process of using AI to draft and review engagement letters. It covers the sequence of using AI for drafting, including gathering matter details, setting scope and fee terms, drafting standard clauses with Playbooks, checking against applicable rules, and lawyer review. It also describes how AI is used for reviewing incoming outside counsel letters against company guidelines, extracting key terms like fees, rates, staffing, and liability, and flagging deviations. The post highlights the importance of scope and change-of-scope clauses, and fees and billing terms in engagement letters.
  3. Can AI Draft Discovery Requests?This post details the technical approach to grounding AI-drafted discovery requests. It outlines the 'Four Anchors' (operative pleadings, client's facts/documents, governing rules/proportionality, prior surviving requests) that ensure defensibility. It also discusses the implications of Rule 26(g) certification, emphasizing the need for 'reasonable inquiry' and attorney verification. The post highlights how Harvey's platform facilitates this grounding by connecting to client documents (Vault), prior requests (Playbooks/Knowledge), and integrating with workflows within Microsoft Word, enabling agents to plan tasks and return cited drafts.
  4. Drafting Clauses in Legal Documents That Hold Up Under PressureThis post details how the AI Demand Letter Generation capability is evolving to incorporate a more judgment-driven approach to clause drafting. It highlights the importance of grounding AI-generated clauses in an organization's specific precedent and standard terms, rather than relying on generic language from the open web. The post also emphasizes the adversarial aspect of drafting, explaining how the system helps lawyers anticipate counterparty interpretations and identify ambiguities. Finally, it outlines a five-step workflow that integrates AI drafting with lawyer review, focusing on starting from an approved position, fast first-draft generation, comparison against standard terms, stress-testing language, and routing to a qualified lawyer for final evaluation. This represents a shift towards using AI as a tool to augment lawyer judgment and improve the quality and consistency of legal drafting.
  5. What Legal Workflows Can In-House Teams Automate?This post details the automation of routine agreement generation, specifically NDAs and DPAs, using AI. It describes how the system leverages an organization's own templates and prior executed agreements to generate first drafts in the company's house language, incorporating established positions. The contribution focuses on narrowing legal involvement to exceptions and enabling business users to generate these first drafts, thereby improving consistency and efficiency for low-variance, high-volume agreements.
  7. AI Legal Brief Writing: Faster Drafts, Stronger ArgumentsThis post details how Harvey's AI assists litigators in drafting briefs and motions faster and building stronger arguments, grounded in citable authority, with every citation verified by the lawyer. It emphasizes the distinction between general-purpose chatbots and grounded legal AI, highlighting the importance of verifiable sources, confidentiality, and the attorney's role in verification, judgment, strategy, and accountability. Best practices for using AI on briefs are provided, focusing on verification, grounding drafts in matter-specific information, using secure platforms, treating output as a first pass, and building verification into the workflow. Harvey supports litigation drafting by generating grounded first drafts of briefs and motions from authoritative primary law, tying arguments to real sources with traceable citations. Drafting occurs directly in Microsoft Word, allowing attorneys to draw on firm precedent and approved language. Custom agents apply firm-specific standards and drafting guidelines consistently across matters, amplifying legal judgment rather than replacing it.
  8. How to Write a Demand Letter Using AIThis post details how to effectively write a demand letter using AI, focusing on prompt engineering, understanding the anatomy of a demand letter, and integrating AI into a reviewed workflow. It emphasizes providing specific facts, governing law, and desired remedies to the AI, and highlights the critical role of lawyer verification for accuracy and tone. The practices outlined include standardizing approved templates, routing drafts through a named review step, and confirming the tool's data handling. The post also categorizes use cases for AI demand letter generation, such as collections, breach of contract, insurance demands, IP disputes, and employment/severance demands.
  9. AI Employment Contract Review: What to Look ForThis post details how AI assists in employment contract review by extracting key terms, checking against playbooks, and flagging exceptions. It outlines specific provisions to check (Compensation, Equity, Benefits; Restrictive Covenants; IP Assignment; Term, Termination, Severance; Arbitration, Governing Law) and patterns to flag (broad restrictive covenants, vague termination triggers, overbroad IP assignment, inconsistent compensation terms, missing protections, jurisdiction mismatches). It emphasizes that AI handles the first pass for speed and consistency, while lawyers retain judgment on enforceability and strategy. Harvey's platform supports this with source-backed extraction, playbook integration, Word/document system workflow, bulk review, and security.
  10. What to Know Before Adopting AI for Case Law ResearchThis post details the architectural choices behind AI case law research tools, focusing on Retrieval-Augmented Generation (RAG) as the pattern for grounding AI output in actual authorities. It outlines five key questions for evaluating AI case law research tools: surfacing adverse authority, jurisdictional precision, tracking treatment and history, transparent reasoning, and accurate pin cites. The post also discusses the evolving role of junior associates, emphasizing verification and judgment skills, and the shift from standalone research tools to integrated AI platforms.
  11. How Lawyers Use AI for Contract DraftingThis post details how AI contract drafting, including generation and revision, leverages retrieval-augmented generation (RAG) with firm-specific templates and precedents. It highlights integration patterns within document editors (Word) and email clients (Outlook), emphasizing the importance of grounding AI outputs in institutional knowledge and the shift in lawyer focus from composition to review and judgment. The capability now supports faster turnaround on repeatable work, higher draft quality from day one, better sharing of institutional knowledge, more lawyer time on judgment work, and increased capacity without headcount.
  12. How Legal Drafting AI is Changing the Way Lawyers WorkThis post details the technical evolution of AI legal drafting, focusing on the grounding layer, retrieval-augmented generation (RAG), and the verification loop. It explains how language models are now grounded in legal sources and the organization's own documents to produce first drafts. The post outlines the workflow sequence: supplying case files/prompts, drafting against retrieved sources, surfacing citations and reasoning, lawyer verification and editing, and final approval. It highlights the importance of the verification loop, including direct citation links and flagging unsupported statements. The post also discusses how AI fits into different stages of the legal drafting lifecycle (blank page drafting, brief building, summarizing dense files, editing/localizing) and provides criteria for evaluating AI legal drafting tools, emphasizing grounding sources, jurisdictional awareness, citation handling, integration, confidentiality, and auditability. It contrasts consumer-grade chatbots with professional-grade platforms designed for legal work.
  13. How to Automate Contract Analysis With AIThis post details the technical implementation of grounding AI outputs in verifiable source language using retrieval-augmented generation (RAG), emphasizing structural comprehension over keyword search, and ensuring domain-specific legal reasoning and verifiable citations. It also outlines practical steps for implementing AI-powered contract analysis, including workflow selection and platform evaluation criteria like domain specificity, verifiable citations, security, and workflow integration.
  14. Introducing the Next Version of AssistantThis post introduces a revamped version of Assistant with three major enhancements: 1. Specialized modes (Assist and Draft) tailored to different use-cases, leveraging different algorithms and models for specific tasks. 2. Iterative functionality for refining and expanding on initial responses, including free-form follow-up questions in Assist Mode and revisions in Draft Mode with 'Show Edits' for tracking changes. 3. Overarching performance improvements, including a 60% reduction in hallucinations, a 23% improvement in cited source accuracy, an 80% improvement in time to first word, better Excel support, enhanced language fluency, and improved prompt understanding for multiple requests. The underlying advancements are attributed to significant progress in agents, retrieval techniques, and evaluation methods, powered by the latest generation of foundation models.