HeyNews

HeyNews drafts newsletters in a publisher's own voice using AI trained on their past issues.

AI / ML · SaaS Live product İzmir, Türkiye
6
0
Oct 25, 2026

The Problem

Newsletter publishers who write on a regular schedule spend hours each issue researching stories and writing a first draft. One customer reports going from 1 hour per day to 5 minutes per day on content creation, and another says HeyNews saved 2-3 hours per newsletter. Finding good stories across RSS feeds, sites, subreddits, YouTube channels, and social profiles also eats hours of a publisher's week. Rushing to hit send often makes drafts sound generic rather than like the publisher's own voice.

The Solution

HeyNews connects to a publisher's newsletter platform (beehiiv, Kit, or a public archive on Substack, Ghost, Mailchimp, Medium, and similar) and learns their voice from past issues, picking up phrasing, tone, and structure. Users follow their sources, and HeyNews automatically checks them and ranks new stories by fit, so the best matches surface first. The publisher picks stories, HeyNews writes a draft in their voice, and the user can rewrite sections, change tone, or try new subject lines in an editor; Style Rules and Reusable Links let users set preferences and approved links for every draft. HeyNews can create finished drafts directly as unsent post drafts in beehiiv (with a paid plan and approved draft access), while on all other platforms the user copies the draft in and sends it themselves. The product never schedules or sends email itself; the publisher always reviews and sends from their own platform.

HeyNews