If you do content — social media, personal brand, marketing — your day is a loop: spot trends, pick topics, write, publish, check the numbers, repeat. Most "AI writing tools" only help with one slice of that. Either they write but can't publish, or they publish but can't analyze, or you have to hand-configure every platform.
Easel (938 stars, Apache-2.0, by the Zhejiang University REAL Lab and Peking University OpenDCAI Lab) aims for the whole loop, end to end.
The five-step loop it automates
- Discover — catch what's trending;
- Plan — turn trends into topics;
- Create — generate image-and-text content;
- Publish — push to Xiaohongshu, Douyin, Zhihu, Bilibili, and more;
- Review — analyze what worked and learn for the next round.
That last step is the difference. It's not a "generate and forget" tool — it's a "publish → measure → learn → do better next time" system. Most AI copywriting tools write for you; Easel operates for you.
Signals it's serious
- Built by top academic labs — Zhejiang University REAL Lab + Peking University OpenDCAI Lab, not a marketing-driven toy;
- 112 skills — the content workflow is decomposed into composable capability modules, not "one big model, one hardcoded prompt";
- Multi-platform — Xiaohongshu, Douyin, Zhihu, and Bilibili have wildly different tones and formats; scheduling all of them uniformly is real engineering;
- Continuously evolving — review data, learn feedback, optimize strategy.
The honest caveat
It's an agent framework, not a plug-and-play app. You configure platform accounts and publish permissions, understand its skill system, and get your hands dirty before it genuinely runs your social operation. If you expect "one sentence → ten posted Xiaohongshu posts," it'll under-deliver. But if you want to build your own controllable, continuously-evolving social media automation system, this is close to the reference answer for "AI social media automation."
I've localized the core docs to Chinese: https://github.com/yangshun2005/Easel-cn
If you find this project useful, a star on the original repo supports the author's ongoing maintenance.
Top comments (0)