Turn Engineering Changelogs into Customer Release Notes Users Read

Paste a changelog or ticket list and get customer-ready release notes grouped into New, Improved and Fixed, with internal-only changes removed and listed.

You getHeadline, intro and new

Prompt

Prompt
Act as a product marketing manager who writes customer-facing release notes for a software product.

I will paste a changelog, commit list or ticket titles right after this prompt. Turn it into release notes that customers will actually read.

Rules:
1. Rewrite every item as a user benefit. Start with what the user can now do or what got better, not what the code changed. No ticket IDs, file names, function names or internal jargon.
2. Drop internal-only changes (refactors, dependency bumps, CI, test fixes, logging, admin tooling) from the customer notes, but list them separately so I can confirm each cut.
3. Group items by impact: New, Improved, Fixed. Put the most visible change first in each group.
4. Keep each item to one or two sentences, under 30 words, written to the reader as "you".
5. If a ticket is too vague to explain to a customer, do not guess. Flag it as a question instead.

Output format:
- Headline: one line naming the biggest change
- Intro: two sentences maximum
- New, Improved, Fixed: bulleted lists under each heading (skip empty ones)
- Action needed: only if a change requires users to do something
- Removed from customer notes: table with columns Item, Reason
- Questions for the team: numbered list

End with one line stating any assumptions you made about who the users are or which changes they will notice.

Example output

At a glanceTallybook 4.12 changes by sectionHow the 14 pasted Jira tickets were sorted into release notes
  • New: 2
  • Improved: 3
  • Fixed: 3
  • Action needed: 1
  • Removed as internal: 5

Example for: Tallybook, a fictional invoicing app for freelancers, version 4.12 with 14 pasted Jira tickets.

Headline

Send invoices in your client's currency, with exchange rates filled in for you.

Intro

This release makes it easier to bill international clients and get paid faster. We also fixed several issues with PDF exports and reminders.

New
  • Multi-currency invoices: Choose from 32 currencies when you create an invoice. Tallybook fills in the day's exchange rate, and you can edit it.
  • Partial payments: Record a deposit or split payment, and the invoice shows the remaining balance automatically.
Improved
  • Faster dashboard: Your dashboard now loads in about half the time, even with hundreds of invoices.
  • Smarter reminders: You can now set reminders for 3, 7 or 14 days after the due date instead of only 7.
  • Client search: Find clients by email or company name, not just by contact name.
Fixed
  • PDF invoices no longer cut off long line-item descriptions.
  • Payment reminders stopped sending twice for some invoices.
  • Dates now display in your chosen format on exported CSV files.
Action needed
  • If you use the Zapier integration, reconnect it once from Settings > Integrations to keep invoices syncing.

Removed from customer notes

ItemReason
TB-2201 Migrate invoice service to Postgres 15Internal infrastructure, no visible change
TB-2207 Bump stripe-node to 14.2Dependency update
TB-2210 Fix flaky e2e test on checkoutTest-only change
TB-2214 Add structured logging to reminder workerInternal monitoring
TB-2219 Admin panel: bulk refund toolStaff-only tool

Questions for the team

  1. TB-2198 "Handle edge case in tax calc": which users are affected, and should we mention it under Fixed?
  2. Is the dashboard speed gain consistent enough to state "about half the time" publicly?
  3. Does the Zapier reconnect apply to all plans or only Pro?

Assumption: users are freelancers and small business owners who care about getting paid, not technical details, so performance and infrastructure work was translated or removed.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Forcing each item to start with what the user can now do turns code-speak into benefits customers understand.
  • Listing removed internal changes in a table lets you verify cuts instead of trusting the model silently.
  • Flagging vague tickets as questions stops the model from inventing features or overstating fixes.

Use before each release when you have raw Jira tickets, GitHub commits or a dev changelog. Paste it right after the prompt in ChatGPT, Claude or Gemini. Add a line about your users if they are non-technical.

Common edits

  • Ask for a short version of 3-5 bullets for an in-app popup or email.
  • Add your brand voice, such as "friendly and casual" or "formal for enterprise admins".

Prompt details

By
Prompts-GPT editors
Published

Did this prompt work?

Keep exploring

More in Product Management