How Draftwise Uses Cohere to Improve Legal Contract Search Quality by 30%

Draftwise was founded on a specific observation about the legal profession: while lawyers describe the most rewarding part of their work as creative strategy and client counsel, they spend most of their time on routine contract drafting and document review. James Ding, Co-Founder and CEO of Draftwise, describes attorneys at client organizations as “cut-and-paste drones” — manually locating relevant clauses, adapting past agreements, and managing the friction of repetitive work before they can engage with the substantive parts of their role.

Building a platform to solve this required more than a search box. Legal documents carry high-stakes implications; accuracy, citation, and data security are non-negotiable. When Draftwise designed Smart Draft, its AI-powered contract drafting product, the team needed a retrieval system capable of surfacing semantically relevant content from clients’ proprietary document libraries — not just keyword matches — and a generation layer that could produce contract language with proper source attribution.

Draftwise evaluated large language models available on Microsoft Azure AI Foundry, where the team’s existing relationship with Microsoft pointed them toward Cohere’s model suite. They selected three models: Command for text generation with source citations, Embed for multimodal semantic search across documents including images and tables, and Rerank to ensure the most relevant results surfaced at the top of every query. Together these three models form the RAG pipeline behind Smart Draft, grounding every generated clause in verified source material from the client’s own agreements.

The results were concrete: Draftwise’s internal benchmarks showed a 30% improvement in search result quality after adopting Cohere’s models. On the backend, API call volume tripled in just the first quarter of 2025 — a direct signal of growing adoption and usage intensity among legal teams.

Draftwise’s Chief Customer Officer noted that excitement about AI within the legal profession has reached genuine momentum, with clients now also asking how they can apply AI to their own operations beyond contract drafting. The company’s deployment shows a pattern emerging in professional services: when AI is grounded in proprietary data with proper citations and security controls, it shifts from a productivity tool to a quality multiplier for skilled practitioners.