The product Kollab made by our team has ended public beta and has been officially released.
Kollab is a team collaboration agent product. To put it simply, it allows AI to not just chat, but actually help you do work, such as drawing pictures, handling bugs, running processes, executing tasks on a scheduled basis, and so on.

Use GPT Image 2.0 to generate images
Kollab is connected to GPT Image 2.0 and can generate images directly in the conversation.
Today is World Book Day, and I asked Kollab to draw a picture of “Book Club Across Time and Space”, which turned out well.

Access chat tool
Bots that connect to IM like Telegram, Slack, and Feishu.
For personal use now, I put Telegram on my desktop and ask Kollab Bot directly if I have any questions.
For team use, @Bot is available in Feishu Group to report bugs. This is the most commonly used scenario.
Anyone who has worked in product customer service knows that the most difficult part of bug feedback is not to fix the bug, but to accurately understand the user’s problem and then convey it to the development team. The loss of intermediate information is very large.
What we do now: User feedback, screenshots, links, descriptions, directly copy and paste, and throw it to Kollab Bot in Feishu Group. Agent will automatically analyze the problem, create an Issue on GitHub, and assign it to the corresponding developer.

There is only one step for product operation: copy and paste. Because the Agent has memory, it understands the architecture of our product, the person in charge of each module, and historical bugs, allowing it to make judgments and assignments more reliably.
Later, overseas communities were established, allowing users to collect bugs through their own @Bot.
Skill reuse
Save a good verified process as a Skill and call it with one click next time.
Two examples I’m using:
- Blog to e-book: Enter the blog address, automatically crawl the articles, format them, generate a table of contents, and output EPUB. Multiple articles are packaged into e-books and imported into WeChat for reading.
- Public account cover image: Enter the article content and let Kollab Agent produce a cover with a unified style based on the article content.
Scheduled tasks
Set the command and time, and Kollab will automatically execute it at the right time. The generated results can be written into Notion or pushed directly as a message.
Our team is currently trying to use scheduled tasks to run the blog content: automatically search for keywords → write the first draft → add pictures → manually review and go online.
The process from search to picture matching is fully automated, and colleagues only do the last step of review.
Launched today on Product Hunt
All of the above functions are already available. Today Kollab is online on Product Hunt. If you have an account, please give us an Upvote:
- Open Kollab’s Product Hunt page
- Log in to your Product Hunt account
- Click the Upvote button
Thank you 🙏 If you have product bugs or suggestions, you are welcome to share your feedback with us.