What it looks like
Here's what comes back when you run it. The case below is synthetic, so you can see the shape of the answer.
Writing a proposal takes us about four hours. We want it down to one. We're thinking of having AI draft it from the call transcript.
1. First principles
The actual desired outcome is that a prospect gets an accurate, persuasive proposal soon after the call, without costing you half a day. One hour is a target. It isn't the outcome.
For that to happen:
- The proposal reflects what the prospect actually said they need
- Scope and price are right, and someone you trust has approved them
- It arrives while the prospect is still interested
- The time you save doesn't come back as rework somewhere else
2. Constraint
Four hours is the total. The skill asks where those hours go, because that informs what the solution is. You might be spending them writing the draft, pulling details together from the call, deciding scope and price, or waiting on someone to review.
If most of the time goes to scope and price decisions, an AI draft from the transcript wouldn’t help. That would require collecting judgement. The skill would clarify where your time is actually spent so you fix the right problem.
3. Intervention
- Start here if the hours go to drafting: AI drafts the proposal from the call transcript. You review for accuracy + any other judgement before it goes out.
- If time goes to decision-making about price & scope: write down your scope and price rules, give the document to your team and/or AI, continue to update pricing sheet to remain current.
4. Second-order effects
- Proposals go out faster. That helps if speed was costing you deals. It doesn't guarantee more closed deals, because that depends on lead quality, price, and timing. Watch the time from call to proposal, and your close rate before and after.
- Quality could slip. A draft built from a transcript can pick up things the prospect said in passing and miss things that were unspoken or were between the lines.
- Review could become a formality. A polished draft is easy to skim. Watch for how much time is now spent from a human reviewer and if the fix actually saves time or just moves the work elsewhere.
- More signed work means more onboarding. Watch whether delivery can keep up.
5. Feedback loops
Hypothetical, harmful: polished drafts make the reviewer check less, so more errors reach clients, so the team spends the saved time on fixes and trust in the drafts drops.
Hypothetical, helpful: the reviewer notes what they fix each time, so the draft instructions improve, so there are fewer fixes next time and review gets faster without getting looser.
6. New constraint
There will always be a bottleneck and it will move elsewhere. The next limit might be the review step or maybe close rates are low because incoming deals aren’t qualified. Watch for this since faster proposals don't create demand.
Next action: time your next proposal by step and capture what a reviewer would update as well as how their judgement is used.
You can ask this skill to expand an option provided, help you analyze the cost of investment for the fix and whether the investment is worth it.
How to try it
Give an AI assistant Github repo or upload the ZIP file. Ask it to read SKILL.md and the linked references first. Then send:
Use Systems Decision Map to assess this business problem. Ask me questions if there’s information missing that would change your decision.
Then just dump your thoughts about your process with AI, share an SOP and/or recent samples of the work.
