Spellbook
AI contract drafting and review inside Microsoft Word for transactional lawyers
Last reviewed 2026-06-18
Spellbook is an AI contract drafting and review platform that runs as a Microsoft Word add-in, plus a separate web app for multi-document work, built for transactional and in-house lawyers. It reviews contracts and flags risky or missing terms, drafts clauses and documents from templates and precedent, answers questions about a contract with linked citations, and benchmarks terms against market standards. Organizational standards are encoded as reusable "Playbooks." In late 2025 Spellbook launched Spellbook Associate, an agentic feature that plans and executes multi-document legal matters: it breaks a goal into tasks, fetches precedents, drafts clauses, and assembles document packages with stated human oversight. As with all legal AI, output requires lawyer review and carries hallucination risk; vendor framing such as "like a junior lawyer" is marketing, not an independently verified claim.
What it can do
Review contracts and flag risk
CopilotRedlines inbound contracts, catches aggressive or missing terms, and surfaces issues against playbook standards; the lawyer accepts or rejects every suggestion.
sourceDraft clauses and documents
CopilotGenerates new clauses or full sections from templates and precedent directly inside Microsoft Word.
sourceAnswer contract questions with citations
AssistantAnswers questions about a contract or relevant context with citations linked to sources in the document.
sourceExecute multi-document projects (Associate)
SupervisedPlans tasks, fetches precedents, drafts, and updates terms across document sets such as data rooms and financing docs, with human oversight at each stage.
source
Strengths
- +Native inside Microsoft Word, where transactional lawyers already work, so there is low workflow friction
- +Multi-model (GPT plus Claude) with citation-linked answers and playbook-encoded standards
- +Spellbook Associate extends from single-document copilot to supervised multi-document agentic workflows
Limitations
- −Output requires lawyer review; like all legal LLM tools it carries hallucination and wrong-citation risk
- −Pricing is opaque (demo-gated), with reported enterprise rates high and rising and seat minimums
- −Agentic claims such as "like a junior lawyer" are vendor marketing and not independently benchmarked
Overview
Spellbook is an AI contract drafting and review tool that runs as a Microsoft Word add-in for transactional and in-house lawyers, with a separate web app for multi-document work. It reviews and drafts contracts, answers questions with citations, and benchmarks terms against market norms.
What it does
The core product reviews inbound contracts against playbook standards, drafts clauses and documents from precedent, and answers questions with linked citations, all with the lawyer accepting or rejecting suggestions. Spellbook Associate, launched in late 2025, is an agentic feature that plans and executes multi-document matters (data rooms, financing, disclosure schedules) with human oversight.
Integrations & setup
Microsoft Word is the primary surface, with Slack, Salesforce, and an iManage document integration. No public developer API or MCP server was documented as of this review.
Pricing
Per-seat subscription, demo-gated. Spellbook does not publish official per-seat numbers; third-party figures circulating online are estimates, not vendor-confirmed.
Best for / not for
Best for transactional and in-house legal teams that live in Word and want review and drafting assistance plus emerging multi-document automation. Not a substitute for legal judgment: every output needs lawyer review.
Alternatives
Robin AI was a comparable contract copilot; Harvey targets broader legal work at large firms.
What people are saying
We aggregate real LinkedIn discussion into sentiment for the agents people search most. Spellbook isn't tracked yet, want it added? Request tracking.
FAQ
Is Spellbook autonomous?+
The core product is a copilot: a lawyer reviews and accepts or rejects every suggestion inside Word. Spellbook Associate runs multi-step, multi-document work but with stated human oversight, so it operates as a supervised agent rather than autonomously.
What models power Spellbook?+
Spellbook describes itself as multi-model, stating it uses GPT and Claude (Opus) among other LLMs, choosing best-in-class models per task.
Sources
- Spellbook (official site) · accessed 2026-06-18
- Spellbook Associate · accessed 2026-06-18
- Spellbook pricing · accessed 2026-06-18
- Spellbook secures Series A (BetaKit) · accessed 2026-06-18
- Spellbook partners with iManage (LegalTechTalk) · accessed 2026-06-18
Last reviewed 2026-06-18