ChatGPT prompt for turning a proposal skeleton into prose

This is the prompt we run once a proposal skeleton is approved, to expand it into full prose without starting from scratch. Its core rule: it keeps every fact exactly as it is, strips the filler, and fills in none of the gaps you deliberately left open.

The prompt to copy

Expand this approved proposal skeleton into full prose in our voice. Rules: benefits before features, their name more often than ours, no filler ("we are excited to"), short paragraphs, one page per section max. Keep all facts exactly as given. Skeleton: {paste}

Replace {paste} with your approved skeleton exactly as it is: bullets, shorthand, and any cells still marked pending included.

When to use it

Once the skeleton is reviewed and approved, when the structure is right and all that is left is the writing. It is the second link in a short chain: an earlier prompt pulls the skeleton out of your discovery notes, this one turns it into prose. You keep control of the structure, the AI handles the tedious writing.

What you get back

1. Full prose in your voice, ready to review.
2. Benefits placed before features, the prospect’s name before yours.
3. Short paragraphs, without the filler phrases that bloat a proposal.
4. Every fact intact, including the cells you left pending on purpose.

Real example: approved skeleton → proposal prose

Input: the skeleton Maya approved for Fernhill Coffee, an artisan roaster the Brightside agency is helping with a range refresh. An excerpt, as-is:

Phase 2 — redesign the 4 signature-range labels (approved on the June 12 call)
Price: [confirm with Lena]
we are excited to join you on this wonderful coffee journey

Output, the part that matters:

Phase 2: your signature range The redesign covers the four labels in your signature range. You gain one coherent identity from product to product, where the current designs were drawn at different times.

Investment: [price TBC]

Notice what the AI did not do: it invented no price for the pending cell, it did not turn “four labels” into a different number, and it cut “we are excited to join you on this wonderful coffee journey”. That is the prompt working, not failing. The human checkpoint, setting the price with Lena, stays in your hands.

The full worked example (complete skeleton, prose generated right after) ships inside the kit, with the discovery prompt that produces the skeleton upstream.