In-house contract work runs on eight types of intelligence. Two arrive from the outside world. Two the department already holds. One has to be written by its lawyers. And three vanish as the work happens unless they're caught. That last group is the whole conversation.
Every type of intelligence the work depends on is in one of four states. The state tells you what it costs to have it. The full map names the same states in data terms, one level more precise.
Kept current by the outside world. Nothing here is unique to any one company, and none of it needs a lawyer's time to maintain.
Everything the company ever agreed sits locked in documents until it's extracted into data. Signing authority sits scattered across systems and has to be read fresh from the systems that own it, never trusted stale, so a lapsed approval is never relied on as current.
The playbook is legal judgment written down: what the department accepts, where it falls back, when it walks. A playbook stored as prose in a PDF is not yet data. Systems can draft positions from redline history; deciding what the department stands behind is the part only its lawyers can do.
These exist for a moment, in a thread or a head. Either they're caught then, or someone excavates email months later, at cost, without what was said out loud, and without a record anyone can trust. They are also the three types every intelligent step depends on most.
The work itself produces the missing intelligence. It has nowhere to land, so it evaporates. Each fix is a field or an event on a surface the team already has. The field is the cheap part; getting it filled under deal pressure is the real build.
Known on day one, gone by drafting. The rationale, the stakes, and what the business needs from it live in an email thread.
Every negotiation teaches the team how this party behaves and what the terms are worth. Today that lesson retires with the deal team.
The approval survives; the reasoning doesn't. So the same exception gets re-argued from scratch next quarter.
Disputes, renewals lost, clauses that failed in practice. The one feed that would make every playbook and every renewal call better, and nobody keeps it.
Positions with their reasons, versioned and owned. And a lawyer's call on whether each outcome was good. Systems can draft both; standing behind them is legal judgment, and that stays with the lawyers.
Fields and events on surfaces that already exist: the request, the approval, the renewal. The schema is the easy part; each field also needs an owner, a shared vocabulary, and a reason to fill it in that survives deal pressure.
What may be recorded, who may see it, how long it lives. Records kept light enough to prove diligence without storing the privileged substance. A capture program without these rules is a discovery program.
This is the briefing. The full map, thirteen steps, eight types of intelligence, and the data underneath each one, is built for the working teams.
Open the full map