dormakaba

dormakaba's legal team — just 9 lawyers (7.8 FTE) plus 3 contract managers — had to handle more than 10,000 contracts per year across every continent, jurisdiction, and language, from simple NDAs to complex SaaS agreements. Experienced lawyers' review judgment was hard to scale ('I know it when I see it'), and manually building clause playbooks in Excel took weeks to months per contract type, with the process repeating for every new contract type and business unit.

Working with legal AI vendor Legartis, dormakaba rolled out a three-phase agentic system: a Playbook Agent captured lawyers' implicit contract-review knowledge through dialogue, Playbook Creation Agents converted that knowledge into full contract playbooks (clause definitions, mandatory/optional classification, risk prioritization) in hours instead of weeks, and an autonomous review layer validated its own accuracy against test cases using F-scores so business units could review contracts independently with legal oversight running in the background.

Review time on 100-page contracts fell 67%, from 1.5 hours to 30 minutes, letting lawyers focus on high-risk clauses instead of routine ones. The system reached an F-score of 0.95 (95% precision), and playbook creation that used to take weeks to months now takes hours. Contract review was delegated globally to business units and junior staff working from the playbooks, freeing the legal team to shift toward strategic work.

• Agentic AI can formalize lawyers' implicit review judgment into scalable, auditable playbooks rather than tacit "gut feel" knowledge.
• Self-validating AI (F-score testing against real cases) builds the confidence needed to delegate contract review to non-legal business units.
• Successful adoption needs an internal "driver" to champion the rollout — the technology alone doesn't produce the result.