Put your release notes on autopilot

Save $29k and 456 hours a year on preparing release notes — while every customer hears about new features and the status of their requests.

Reads your tracker

Posts to your channels

You shipped it.

Your customers never heard about it.

So they never use it.

80%

of features are rarely or never used.

Every release needs a note, in every channel they read.

Customers wait for a fix you already shipped

Support hears about it from the customer

Leadership sees tickets, not value

Ducalis saves you $29,434
and 456 hours a year
on preparing release notes

By hand $32,760288 versions × 1 h 45 min × $65/h. A version is one release note for one audience, in one channel, in one language; versions per release = channels × languages. Rate: Indeed, US product manager average base salary $135,269 a year (July 2026) ÷ 2,080 working hours.Learn more →/ year
With Ducalis AI $2062 seats × $96 + 48 releases × $0.30 AI. Seats: Ducalis licence, $96 per user a year on the annual plan. AI: about $0.20 of tokens per release plus $0.02 per extra version, measured on DeepSeek V4 Flash with your own key. Your review, about 10 minutes per version, is counted in the hours, not in this bill./ year

Set up once. Get release notes on schedule.

  1. Write one prompt

    Tick what you need below and copy it.

  2. Put it on a schedule

    Every Friday, every Monday or after each release.

  3. Release notes come on their own

    Your team gets them as is. Customers get a draft you send.

Goal
When Tracker Send to

Your team gets the full list in Slack. Customers get a release-note draft you send. Every Friday at 10:00 — after a test run you approve.

Make this a routine: every Friday at 10:00 [timezone], collect what shipped in Linear since the last run, then: 1) post the release notes for the team to [channel] in Slack; 2) add the customer-facing changes to a release-note draft on [board name]. Do a test run first; turn on the schedule after I approve.

What happens on every run

// in

Linear Jira GitHub Voting board Slack threads
┌─ READY FOR YOUR REVIEW ────────┐
│  Email draft                ✓  │
│  Changelog                  ✓  │
│  4 ideas in Done            ✓  │
│  #product-updates           ✓  │
├────────────────────────────────┤
│  Draft: will not be sent       │
│  You review, it ships          │
└────────────────────────────────┘

// out

Customer email Changelog Ideas in Done Team Slack Leadership digest

// every channel

One release note. Everywhere your people read.

Write it once. Ducalis delivers it to every channel, worded for who reads it.

Open our changelog widget

Acme

What’s new — October 2026

Here’s what we shipped in Acme Billing over the last two weeks.

Recurring invoices are live

Set an invoice to repeat monthly — we send it on schedule.

Related request: Recurring invoices

Export invoices as CSV

Send your invoices straight to your accountant.

Related request: Export invoices as CSV

View all changes

// every audience

Ducalis AI rewrites each release note for who reads it.

Public or internal, with bugs or without, any language. Ready out of the box — nothing to set up.

What’s new in Acme Billing

A

Acme Billingto you

10:02

Recurring invoices are live. Set an invoice to repeat monthly — we send it on schedule.

Export invoices as CSV. Send them straight to your accountant.

See what’s new
// close the loop

Tell every customer their request is done.

Each person gets a release note about what they asked for — not a digest of everything.

Maya asked for CSV export

CSV export is live

Leo asked for custom roles

Custom roles are live

Ana reported: PDF won’t download

PDF download is fixed

// every version

Every deploy ships its own release note.

Your CI calls one webhook after each deploy. The release note goes out on its own.

  • Jenkins
  • Bamboo
  • GitLab CI
  • GitHub Actions

Release webhook

  • Release note sent
  • Changelog updated
  • Ideas → Done, voters told

Same release notes from your own AI.

Everything the agent does in Ducalis chat works from Claude, ChatGPT and Codex via MCP — the same prompts, routines and channels.

  • Ducalis AI chat
  • Claude
  • ChatGPT
  • Codex
Connect via MCP →

Build vs buy

Sure, you can vibe-code a release-notes bot. The question is how long it takes, how reliably it runs, who keeps it alive — and whether that’s the best use of your time.

Vibe-code it yourself

2–3 days

to build and test, then 4–11 h a month to keep it aliveTracker APIs, tokens and models change: in 2025–26 Jira removed its old search API, Atlassian tokens started expiring yearly and Slack retired its upload API.

  • Tracker API and pagination
  • Link issues to customer requests
  • A prompt per audience
  • Slack bot, email, changelog page
  • Cron, secrets, retries

Use Ducalis Free

$0

up to 500 subscribers — see pricing

  • Connect your tracker in the chat
  • Votes and requests already on your board
  • The agent writes each version
  • Email, changelog, Slack built in
  • Routines with a run log
Start free

One Ducalis run

$0.30

per release note

7 shipped issues → 12 versions · under 10 minutes

Finds

  • Triggered by your deploy
  • What shipped in your task tracker
  • Matches it to customer requests
  • Privacy check before anything goes out

Writes

  • Release note for customers
  • Team notes, with bugs
  • Leadership summary
  • An email to every voter
  • In your customers’ languages

Delivers

  • Email
  • Changelog page
  • In-app widget
  • Slack
  • Telegram
  • Discord
Estimated from measured runs on DeepSeek V4 Flash via OpenRouter, 29–30 Oct 2026: a full release thread cost $0.17. The demo release: 7 issues from Jira and Linear, matched to 38 customer requests, written for 4 audiences in English, Spanish and Japanese (12 versions), posted to six channels; rounded up to cover it. Voter emails, the changelog page and the widget use no AI tokens. AI cost depends on your model and volume; you pay your provider directly.

Release-notes tools compared

Feature DucalisLaunch
Notes
Feature
base
BeamerAnnounce
Kit
Linear
Drafts from your trackerAI reads the work you shipped — no copy-paste. ✓Linear, Jira and 20+ trackers. ✓Jira filter or JQL, plus Confluence and files. ◆Only from requests synced from Jira or Linear. ✗You type a short description. ✗You type a title and keywords. ✓Linear issues in a release pipeline only.
Knows who askedLinks each change to the customers who voted for it. ✓Ideas, votes and requests on your voting board. ◆Feedback management in the Growth plan. ✓Drafts from closed feature requests. ◆Voting is a $99 add-on. ◆Feature requests are a separate $49 product. ✗No customer voting.
Your own AI keyPick any model and pay the provider directly. ✓Any model with tool calling, via OpenRouter or your key. ✗No — “bring your own” covers only the email provider. ✗Not offered on the pricing page. ✗Not offered on the pricing page. ✗Not offered on the pricing page. ✗Not offered on the pricing page.
Team post in SlackThe team version goes to your channel. ✓Slack, Telegram, Discord, Mattermost and more. ✓Slack integration. ✓Slack in the Professional plan. ✗Widget, page, push and email only. ✓Slack integration. ✗Stays inside Linear.
Public changelog + widgetCustomers read updates inside your product. ✓Changelog page and in-product widget. ✓Page and in-app widget. ✓Page and in-app widget. ✓Widget and standalone page. ✓Widgets. ✗No public changelog.
AI from, per monthCheapest plan with AI release-note drafting, team of 5. $0Free plan with your own model key; paid plans from $10 per user. $799Growth, billed yearly; 2 users. $295Professional: $59 per seat × 5. $99Pro, billed yearly. No AI on Starter. $98Changelog $49 + email add-on $49, billed yearly. $80Business: $16 per user × 5.
Cheapest plan with AI release-note drafting for a team of 5. Source: each platform’s official website and pricing page, checked on 29 Oct 2026. Vendors change plans without notice.

Questions

Can I generate release notes from Jira, Linear or GitHub?

Yes. Connect your tracker in the Ducalis chat and ask what shipped. The agent reads completed work and links it to ideas on your voting board.

Does it send anything without me?

No. The agent prepares drafts; you press Send. Anything that would make internal work public waits for your OK.

Can customers and my team get different versions?

Yes. The agent asks who it’s for. Customers get the benefits; your team gets every fix and what to tell customers.

Which AI model does it use?

Any model with tool calling — through OpenRouter or your own key. Our full demo run cost $0.13 on DeepSeek V4 Flash.

I already use Claude or ChatGPT. Why Ducalis?

Keep them — connect through MCP. Ducalis adds your tracker, customer votes and delivery in one place, plus routines that run on schedule.

Sign up. In 10 minutes your release notes are on autopilot.

Start free

No credit card required