Spellbook raises $50M to scale contract review and move deeper into transactional work
Legal AI platform Spellbook closed a $50 million Series B led by Keith Rabois at Khosla Ventures, valuing the company at $350 million post-money. New investor Threshold Ventures joined, alongside Inovia Capital, Bling Capital, Moxxie Ventures, Path Ventures and Jean-Michel Lemieux. Total funding now tops $80 million.
What Spellbook offers
Launched in 2022, Spellbook is a generative AI tool built for lawyers with a focus on contract development, accuracy and improvement. The company says its system uses OpenAI's GPT-5 with legal-specific methods to meet contract law expectations. "We're at the spreadsheet moment for lawyers," said co-founder and CEO Scott Stevenson. "Just as spreadsheets transformed accounting, large language models are transforming law after 20 years of technological stagnation."
Spellbook integrates inside Microsoft Word as a drafting and review companion. Attorneys get AI suggestions and feedback, access a clause library, and apply playbooks with rules and recommendations by contract type.
It can connect to OneDrive, Dropbox or direct uploads to ingest a firm's prior agreements. That enables teams to locate and reuse proven clauses in context, instead of keyword hunting across folders.
Scale and customers
Spellbook reports more than 10 million contracts reviewed on the platform. Customers include Nestlé, eBay and Kennedys Law. The company says it now serves nearly 4,000 law firms and in-house teams across 80 countries.
Agentic review with "Associate"
Similar to coding assistant Cursor, Spellbook's Associate can review multiple documents, update terms and party details, and flag or fix issues. Lawyers work with it through a chat interface to query language and apply revisions across a document set in one pass. Throughout, the agent takes attorney feedback to match revisions to guidance.
Reducing "AI slop"
Spellbook says it grounds outputs in real-time information and data to help cut down on low-quality or incorrect content. The goal: keep attorneys in control while improving speed on routine contract cycles.
Risk context you should note
General-purpose AI has created risk for legal teams due to hallucinated case law and fabricated citations. Morgan & Morgan warned its attorneys earlier this year after a Wyoming judge threatened sanctions over fictitious cases. In September, California issued a $10,000 fine for a filing that included fabricated citations, stating 21 of 23 quotes were made up.
What this means for legal teams
- Start with bounded use cases: NDAs, vendor MSAs, SOWs and standard amendments.
- Pilot inside Word with a narrow playbook and a short, known clause library.
- Connect prior agreements (OneDrive/Dropbox) to enable clause reuse and reduce variance.
- Keep human-in-the-loop review mandatory; require source links for any legal authority.
- Track concrete metrics: review cycle time, redlines per draft, approval rate, and error rate.
- Define escalation rules: which edits the AI can propose vs. which require partner sign-off.
- Address privilege and confidentiality upfront; restrict data sharing and log access.
Market direction
With this round, Spellbook plans to expand from contract review into broader transactional work. The company is targeting the $1 trillion transactional legal services market and rolling out Market Comparison and Preference Learning in beta to deepen contract intelligence grounded in market data.
Bottom line
For firms and legal departments, the pitch is clear: keep lawyers steering, give them speed on the hills. If you're evaluating tools, prioritize integrations with your document stack, grounding on your deal history, and controls that prevent fabricated citations.
Want structured upskilling in legal-focused AI workflows? Browse role-based AI learning paths at Complete AI Training.
