Most legal technology asks your team to change before it delivers anything: licences, training, an adoption curve. Flank inverts that. Agents do the work, so implementation means the agents learning your workflows, not your lawyers learning an interface.

Agentic deployments fail on expectation gaps, so we close the gap before kickoff. The Agent Design Document specifies the use cases in scope, the playbook approach, the supervision model and what escalates to a human. You sign it, and so does the forward-deployed lead accountable for delivering it.

Two weeks of discovery and design, two weeks of playbook engineering with your lawyers, then agents go live in week five on a contained scope: one document type, one region, fully supervised. Scale comes after the evidence.

The honest list is short and sits mostly in the first four weeks: a named point of contact, your templates and a sample of recent documents, an hour or two a week from your lawyers, and IT sign-off on the channels the agents work from.

Every workflow launches fully supervised, and supervision sits wherever your risk model wants it: your team, a partner firm, or Flank. Implementation is scoped in forward-deployed engineering days in the design document, and because pricing follows outcomes, paying starts when the system starts delivering.