What's new in 3.0

• Unified inbox — feedback, ideas, and changelog comments in one place. No more switching tabs to see what users are saying.
• Instant publish — go from draft to live in one click. Subscribers receive their email notification in under 60 seconds.
• Project switcher — manage multiple products from a single dashboard without signing in and out.
• Redesigned editor — rich text, inline images, code blocks, and a new AI draft button for generating your first-cut release note from bullet points.

Performance improvements

Public changelog, feedback, and roadmap pages now load in under 800 ms on a cold start — a 3× improvement over v2.x. We migrated to edge-cached static rendering, so every page load is fast regardless of your subscribers' location.

How to upgrade

If you are on a legacy plan, your account will be migrated automatically over the next 48 hours. No action needed. You'll receive a confirmation email when the migration is complete.

New users start on 3.0 immediately.

What's new

• Dark mode — toggle in your profile settings or let it follow your OS preference automatically.
• 5 built-in themes — Midnight, Slate, Forest, Coral, and the default Light. Each theme adjusts sidebar, card backgrounds, and data visualisation colours consistently.
• Custom brand colour — Pro and Agency accounts can now set a custom accent colour that ripples through buttons, active states, and chart highlights.

Why we built it

Over 60% of you voted for dark mode on our ideas board over the past 18 months. It took longer than we hoped because we refused to ship a half-baked toggle that left tables and modals on a white background. This is the real thing.

How to enable it

Go to Settings → Appearance and choose your preferred mode. Changes save instantly — no page reload needed.

What's new

• Active users chart — daily, weekly, and monthly active users on one graph, with a 90-day rolling window.
• Feature adoption heatmap — see which parts of your product are used most and which are gathering dust.
• Retention cohorts — understand whether users who signed up in January are still around in July.
• Churn signals — accounts that haven't logged in for 14+ days are flagged automatically so your CS team can act before they cancel.

All charts are filterable by plan, country, and signup date. Data exports to CSV in one click.

Available notifications

• New feedback submitted (with rating and message preview)
• High-upvote idea threshold crossed (configurable — default 10 votes)
• Changelog published to subscribers
• New subscriber milestone (100, 500, 1000…)
• Churn signal triggered for an account

Setup

Go to Settings → Integrations → Slack and click Connect. You'll be redirected to Slack to authorise the app, then choose which channel receives which notification type. Takes under two minutes.

What changed

• Lazy loading — the widget script no longer blocks your page render. It loads after your app is interactive.
• Mobile-first — the popover is now fully responsive and works correctly on iOS Safari and Android Chrome.
• Theme tokens — control background, text, badge, and border colours via CSS variables. No more overriding internal class names.
• Unread badge — shows the number of new entries since the user last opened the widget, not just a dot.
• Keyboard accessible — opens and closes with Escape, navigates with arrow keys.

Existing widget installs will pick up the new version automatically on your next deploy. No script tag changes needed.

How it works

1. In the changelog editor, click AI Draft.
2. Paste your commit messages, ticket titles, or raw notes.
3. Flowdesk generates a structured, user-friendly entry — categorised, with a title, body, and suggested CTA.
4. Edit, approve, and publish. Or regenerate with a different tone (technical / casual / executive).

The AI never publishes automatically. You always review before anything goes live. Think of it as a first-draft assistant, not an autopilot.

What caused it

A misconfigured retry policy on our email worker was causing failed sends to re-queue at the back of the global job queue, behind unrelated background tasks. Under heavy load, this created a compounding delay.

What we fixed

• Email sends now use a dedicated high-priority queue, isolated from background processing.
• Retry logic has been tightened — failed sends retry 3 times within 5 minutes before raising an alert.
• We added real-time monitoring on email send latency with an automatic alert if median delivery time exceeds 60 seconds.

If you were affected by delayed notifications in April, we're sorry. This is now resolved and monitored continuously.