You take the sales call and promise the scope by end of week.
Case Study 01 / 09
Proposal Pipeline
I run this for my own agency's proposals. Four agents: a planner, a drafter, a checker, a reviewer. A call transcript goes in one end, a finished scope comes out the other, and the first human read is the last pass before it ships. This page is the anatomy.
You re-listen to the recording and pull the scope out by hand.
You open an old proposal and start swapping names.
You check the numbers against what you remember saying on the call.
You proofread your own writing, so the mistakes you can't see stay in.
The proposal goes out whenever you find the hours, and the deal cools while it waits.
You take the call. That part is the business.
A planner agent reads the transcript and maps the scope: phases, deliverables, assumptions.
A drafter agent writes the full proposal in the house format.
A checker agent reads the draft against the transcript and flags anything you never said.
A reviewer agent makes the pass a partner would: tone, pricing logic, holes.
The finished draft is in front of you before the deal cools, and the last read is yours.
The Hand-Off · Four agents in, one read out
Draft promises in phase one. Not in the transcript. Cut or confirm.
-
You
The call. Agents don't take meetings.
-
You
The price. The pipeline drafts a number; setting it is your call.
-
You
The send. Nothing leaves with your name on it that you haven't read.
In a session, this workflow gets one of three verdicts: automate it now, fix it manually first, or leave it alone.
A proposal pipeline earns "automate it now" when the format is already stable and the calls keep coming. If every scope you write is a one-off, it's "fix it manually first": the template gets locked by hand before any agent touches it. And if proposals are rare in your business, leave it alone. A pipeline needs volume, and yours would mostly sit idle.
Sixty minutes. The spec for this, written live.
Book the session