// Case study
AI autoposting for a Telegram channel
Every day the system finds the most engaging AI news from competitors, writes a digest in the channel's voice, draws an image and publishes it after one tap.
// Task
A daily post takes hours of manual work
The owner of an AI channel had to go through dozens of posts in several channels every day, find what actually resonated, rewrite the news in their own voice without copying, and find an image for every post.
// Solution
A pipeline from collection to publishing
At 21:00 the first scenario reads the public pages of source channels, counts views and reactions and filters duplicates. At 21:15 the second one picks the top 5 posts by engagement (ER = reactions ÷ views, so a small channel's viral post doesn't lose to a big one), and OpenAI writes a digest in the channel's style and draws an image.
The draft goes to the owner's bot. Nothing is published without a "yes": approve, regenerate the text or regenerate only the image. The third scenario reacts to the buttons instantly and publishes the post.
- Sources live in Google Sheets: add or disable a channel without touching the scenarios
- Ranked by real audience reaction, not by view count
- In Ukrainian, in the channel's voice, with no copying or made-up facts
- History of every digest with sources, image prompt and status
// Architecture
How it works
// Result
The outcome
You just tap "Approve". The system does the rest.
// Want the same?
Need autoposting?
Send me 10–20 of your posts and a list of sources, and I'll show you what the digests would look like for your channel.