Document Processing · Risk Management
Spellbook needed to review, redline, and route very large volumes of contracts fast enough for lawyers and legal teams, while keeping issue spotting consistent and aligning recommendations with each team's preferred review style.
Reported impact
Spellbook now runs about 530,000 contract reviews and more than 700,000 lawyer chat interactions per month, cuts agreement work from roughly 10 hours to about one, and extends the same Claude stack into autonomous contract-management workflows that move agreements through the organization.
Spellbook runs a mostly-Claude contract workflow on the Claude Platform. Contracts can be pulled in from email, Slack, or Salesforce before a lawyer opens them, then Claude-powered agents flag risks, draft redlines inside Microsoft Word, compare terms to market, and feed newer autonomous contract-management flows that move agreements through triage, review, negotiation, storage, and ongoing monitoring. Roughly 15 Claude agent configurations handle the review surface alone, matching model depth to task complexity: Opus 5 and Opus 4.6 on the hardest subtasks, Sonnet 4.6 where speed and tone matter, and Fable 5 as the oracle model in evaluation pipelines. Spellbook judges production quality by user acceptance of suggestions and repeated consistency checks, while its engineering team uses Claude Code with Fable planning and review to keep shipping new legal workflows.
Contracts arrive from email, Slack, or Salesforce, Spellbook routes them through its Claude review agents to flag issues and draft redlines in Microsoft Word, and the same normalized contract context can continue into triage, negotiation, storage, and monitoring workflows.
Cuts the time to complete an agreement 10x, from roughly 10 hours of lawyer work to about one
Roughly 15 agent configurations just for reviewing contracts
Runs its Claude-powered review agents across about 530,000 contracts every month
Spellbook recently launched autonomous contract management, which moves agreements through an organization
We'll use Fable as an oracle model to generate a set of criteria
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Google Cloud Sample Archived · Sep 17, 2026 Parsewise For risk specialists in the financial services and insurance industries, reading can feel like a full-time job. Whether assessing an insurance claim or approving a loan, these subject matter experts must search through… 29% lower error rate than best-performing LLM on grounded-reasoning benchmark Gemini · Gemini 3.0 Flash · Gemini 3.5 Flash 
Anthropic Sample Archived · Sep 15, 2026 DXC Technology DXC needed to reduce insurance document backlogs and automate highly regulated underwriting and claims decisions without losing auditability, policy fidelity, or human oversight. 8 hours to stand up a regulated benefits calculation, 80% right on the first pass Assure · Claude Code · Claude Platform 
MongoDB Sample Dec 22, 2025 Xlrt Xlrt needed to help a multinational bank overcome data silos and manual contract workflows that slowed forecasting, procurement analysis, and vendor management across many regions. Processes that once took 15 days now take just three, which is 75% faster and more cost efficient. MongoDB Atlas · MongoDB Enterprise Advanced 