Verified against ChatGPT · 2026-07-30
Turn messy bullet notes into a coherent paragraph
Converts scattered meeting notes, brainstorm fragments, or half-sentence bullet points into one or more flowing paragraphs with real transitions — without inventing connections or facts the notes don't support.
The prompt
Ready to copy — highlighted parts are example details you can swap.
TASK: Turn the raw notes below into a single paragraph — clear, connected prose, not a list. Keep every distinct point from the notes; don't drop any and don't add any that aren't implied by what's there.
RAW NOTES:
- customer churned after 3 months
- said onboarding was confusing
- support ticket volume high in month 1
- competitor has better docs
- CS team flagged this in Q2 review
- pricing wasn't the issue per exit survey
CONTEXT (what these notes are for, if it changes how they should read): This is going into a churn analysis summary for the product team, to argue we should invest in onboarding.
REQUIREMENTS:
- Group related notes together into logical paragraphs rather than following the original bullet order if a different order reads more naturally
- Add real transitions between ideas ("as a result," "meanwhile," "this matters because") only where the notes actually support that relationship — don't invent a causal or sequential link that isn't there
- If two notes seem to contradict each other or you're unsure how they connect, flag it rather than silently picking one interpretation
- Match this tone: plain, analytical, internal-memo tone
- Keep it to roughly 100-120 words
Write the paragraph(s). Then list, in one line, any note you had to guess the intended meaning of, so I can confirm or correct it.Customize the highlighted detailsoptional — the prompt above already works
Why this works
Turning fragments into prose is exactly where LLMs are most tempted to over-connect — smoothing choppy notes into flowing paragraphs naturally invites invented causal links ("as a result," "this led to") that the underlying notes never actually established. The explicit instruction to add transitions only where the notes support them, plus a required flag for guessed connections, directly targets that failure mode instead of leaving the model free to narrate a story that sounds coherent but overstates what the notes actually say. Asking for regrouping by logical relationship rather than preserving original bullet order is what makes the output read as genuinely written rather than a bullet list with periods swapped in for dashes — order-preservation is the default and usually the wrong choice for notes captured in the moment rather than in argument order.
What you get back
The customer churned after three months, and the pattern points primarily to onboarding rather than price: the exit survey explicitly ruled out pricing as a factor, while support ticket volume spiked in the first month and the CS team flagged onboarding friction in the Q2 review. Confusing onboarding also compares unfavorably to a competitor whose documentation is stronger, which may be compounding the problem rather than causing it independently. Guessed connection: I inferred that the competitor's better docs are a contributing factor rather than a separate, unrelated note — worth confirming that's the intended read.
Verified against
ChatGPT GPT-5.1 · 2026-07-30
Claude Claude Sonnet 4.5 · 2026-07-30
Gemini Gemini 3 Pro · 2026-07-30
Changelog
- 2026-07-30 — Added the guessed-connection flag requirement after finding models would confidently assert causal links between unrelated notes.
- 2026-07-13 — Initial version published.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
