← All guides

Preconstruction meeting brief: what the estimator actually needs in the room

Precon meetings burn time when the estimator shows up with tribal memory and a laptop full of tabs. The room needs a short brief, not a scavenger hunt through the bid folder.

The job of this page

Name what has to be in the room before the meeting starts. Keep it short enough to read. Own who builds it. Only then use AI to draft from the package you already have.

Why tribal memory fails the meeting

Someone remembers the productivity assumption. Someone else remembers an alternate that might still be open. Risk lives in a side note. Open questions sit in an email from three weeks ago. The meeting becomes a reconstruction exercise instead of a decision room.

Ops hears half the story. Estimating spends the hour defending memory. Nobody leaves with a shared basis.

The short brief that must be in the room

One page if you can. Two if you must. Same spine every time:

• Basis of estimate. What you assumed on units, productivity, labor, and markups. Written so the room is not guessing.
• Open questions. Still unresolved with owner or design. Who owns the chase after the meeting.
• Risk flags. The few things that blow the job if ignored. Named, not “watch this.”
• Alternates and options. What was priced, what is selected, what still needs a decision in the room.

If those four are not on paper (or a shared doc the room can see), you are not ready for precon. You are ready for storytelling.

Where AI helps

AI can draft the brief from the package when the inputs exist: basis notes, risk list, alternates, open questions in a place the model can read. It drafts. A human owns the brief before it hits the room.

AI cannot invent a basis you never wrote down. Dumping the whole bid folder and hoping for a meeting-ready brief is still garbage in.

One concrete picture

Mid-market GC. Precon for a won package. Estimator walks in cold, opens five PDFs, and reconstructs the bid from memory while the PM takes notes. An alternate assumed selected never gets said out loud. A risk flag on access shows up after mobilization.

Same company, next award: a one-page brief is required before precon. Basis, open questions, risk flags, alternates. AI drafts from the structured handoff package. Estimator edits and owns it. The meeting starts on decisions, not archaeology.

Scene is an illustrative pattern, not a named client case study.

What “done” looks like

Every precon has the same short brief in the room, with a named owner, before the meeting starts. AI may draft it. Tribal memory is not the agenda.