What's new: connect once and start writing sooner, wider story window, Content Desk keeps your edits
Hey everyone, posting today's release notes.
Most of what shipped is in three places: onboarding, story access, and Content Desk resilience.
What changed:
Connect once, start writing sooner. When you connect @beehiiv, Kit, or a newsletter archive during setup, onboarding now finishes from that connection, and Voice Training starts right away. If training succeeds, your AI Writer is ready to use without waiting for manual approval. You'll get a note when it's ready.
A wider 12-month story window. Usable accounts now share the same 12-month story access window. Recent older stories no longer disappear behind plan-specific upgrade prompts as long as they're inside the 12-month storage horizon.
Content Desk keeps your edits. Adding a URL story is less brittle. If a page has thin metadata or extraction is partial, HeyNews keeps the title, details, tags, and content you entered. You can also edit story titles and details after saving. Imported stories keep those edits for the current newsletter only, so shared story records don't get rewritten for your other newsletters.
Sources are easier to maintain. Fetch now is back for RSS feeds, social profiles, and shared sources. Source names, descriptions, notes, and tags are saved as source-level metadata rather than overwriting the article or video titles attached to stories.
Plus the usual fixes: ActiveCampaign public archives import the real issue URLs instead of treating archive index pages as issues, fresh Content Desk stories stay visible in Compose before story scoring finishes, removing a source no longer leaves stale source groups or hides the same story from your other newsletters, Voice Training handles larger past-issue libraries more reliably during source discovery, and cancel/save-offer choices are guarded against duplicate clicks and race conditions.
Full release notes: https://roadmap.heynews.co/p/whats-new-may-21-2026-AAbZDf
If you hit anything weird in the new auto-start training flow or with Content Desk edits, reply to this thread, and we'll dig in.
Cagri, co-founder
Replies
The 12-month story access update is a good move. Users get frustrated quickly when older content suddenly disappears behind upgrade prompts.
I like the idea of separating source metadata from story metadata. That feels much cleaner architecturally and probably avoids a lot of accidental overwrites.
Content Desk preserving edits sounds like a small change, but honestly those workflow reliability fixes matter a lot 😅 Losing edits is one of the fastest ways to break trust in writing tools.
The source-level metadata change feels important. For newsletter tools, a “story” is rarely just the URL — it’s the source relationship, why the publisher trusts it, the angle they’re considering, and the edits they made for this issue.
One thing I’d watch in the auto-start Voice Training flow: show a small training receipt. Something like “trained on 43 past issues, ignored 6 because extraction was weak, detected these recurring sections/tones.” That would make the model feel less mysterious and give publishers a way to catch bad source material before it shapes the writer.