Spellbook launches AI contract lifecycle management

Spellbook, a Toronto-based legal-AI company, released Autonomous Contract Management (ACM) to a limited set of early-access customers on June 30, 2026, according to an official Business Wire press release and BetaKit reporting. CEO and co-founder Scott Stevenson told BetaKit the product is the "infrastructure for agreements" - the "invisible thread" tying organizations together - saying, "Behind every data centre, every rocket launch, are thousands of agreements. It's underappreciated." In the official release, Stevenson said: "CLMs were the first attempt at solving a real problem, but they were built before AI could do the actual work ... We built Spellbook from the AI layer up so the work happens in the background, and lawyers wake up to a queue that's already been handled." Spellbook says it serves more than 4,500 legal teams across 80 countries, including Dropbox's in-house legal team and global law firm Kennedys, per BetaKit; the company was most recently valued at $350 million USD in a Khosla Ventures-backed round, per BetaKit.

Per the official release and BetaKit, ACM operates across three workflows: intake (pulling contracts autonomously from email, Slack, and Salesforce, then triaging and redlining them against a team's standards before a lawyer opens them), review (each negotiation in its own workspace with automatic review of counterparty redlines and version comparison), and insight (signed agreements stored and searchable with citations, plus renewal flagging). Stevenson told BetaKit that a feature called Spellbook Radar, launching in Q4, will flag external policy or regulatory changes - for example, a province updating an employment law - that could affect existing contract clauses.

Artificial Lawyer, in independent coverage, frames ACM as effectively a "CLM killer" (its characterization, not the company's), arguing Spellbook's advantage over incumbent contract-lifecycle-management vendors is that AI-driven contract review was always its core product, with the broader platform built outward from that capability rather than bolted onto a document-storage system. Companies offering end-to-end contract automation commonly face three engineering priorities: reliable information extraction across heterogeneous sources, secure integrations with collaboration and e-signature systems, and provable audit logs for compliance and legal defensibility. Artificial Lawyer also flags an open question about token costs at scale, since large customers may want tens or hundreds of thousands of contracts inside the platform, which has agentic, higher-token-use workflows built in.

BetaKit reports Spellbook cites an internal Spellbook Labs study finding 60% of SEC-filed contracts had drafting errors, with 1 in 40 containing a "high-risk" error - the kind of baseline error rate that vendors in this category use to justify automation, though it is a company-produced statistic and has not been independently replicated. Teams evaluating similar tools should ask for audited accuracy metrics on drafting and clause extraction, confirm how access controls and retention policies are implemented, and test change-detection for regulatory or jurisdictional updates before relying on outputs in production.

• •Whether Spellbook expands ACM beyond early access to its full customer base in the one-to-two-month window the company targeted.
• •The Q4 launch of Spellbook Radar and how it sources external regulatory/policy signals.
• •Whether competing contract-AI vendors and legacy CLM platforms respond with similar end-to-end offerings.
• •Independent, audited accuracy data on ACM's drafting and redlining output as it scales past early access.

The launch is corroborated by Spellbook's own release, BetaKit's on-the-record interview with Stevenson, and independent trade coverage from Artificial Lawyer, so the core facts - product scope, timing, and customer base - are solid. The company-produced statistics (4,500+ customers, the 60% SEC-contract-error study, the $350 million valuation) are all Spellbook's own figures and should be read as vendor claims rather than third-party-audited numbers.