Spellbook - AI Contract Drafting & Review

Review is the core feature, and it works on a full document rather than one clause at a time. Spellbook scans for client risks, drafting errors, and missing terms, then drafts comments and redline suggestions you can accept or reject. Edits appear under your own name, so you can forward the marked up file to a client without extra cleanup. Bulk approve lets you accept several changes in one go.

A playbook saves a set of instructions for how you want documents reviewed, so Spellbook checks them against your positions instead of a generic checklist. Firms can reuse the same playbook across every matter, which keeps review consistent when several lawyers touch a deal. You can save and reuse specific instructions and adapt them over time.

Draft writes new clauses and documents, or pulls language from your own precedent collection. The tool reads the substance of the open document and proposes ready to use text that fits it. A clause library stores your favourite clauses as reusable building blocks, and inserted language lands in line with one click instead of a copy and paste round trip.

Associate is an AI agent that handles projects spanning several documents. It breaks a goal into tasks and runs them, updating terms, party details, and issues across a document set in one click. It fits heavy packages like dataroom reviews, financing documents, disclosure schedules, and employment bundles, and it tracks connections between documents so accuracy holds across the whole set.

Compare benchmarks your contract against thousands of similar agreements, filtered by industry, jurisdiction, and deal type. You can see how common a term is and what alternatives exist in the market, with percentage breakdowns for each. Ask handles legal research over 1,000 vetted statutes, regulations, and government databases and returns answers with the underlying source cited.

Spellbook runs on Amazon Web Services and holds SOC 2 Type II, HIPAA, GDPR, and EU AI Act compliance claims. It has zero data retention agreements with OpenAI and Anthropic, so customer data is not used to train third party models. It supports single sign on and stores customer data in data centers in Canada and the United States, and it works in Microsoft Word and Google Docs.