Colleague Skill: China's Viral AI Tool That Clones Fired Employees' Knowledge

Training AI to replace your coworker

An open-source project called Colleague Skill has taken China by storm. The tool, designed to transfer a worker's professional knowledge to an AI agent before their departure, is being widely discussed on the social platform RedNote (known domestically as Xiaohongshu) and has amassed over 8,000 stars on GitHub.

The developer positions Colleague Skill as a more efficient way to "say goodbye" to colleagues. The system absorbs an employee's professional expertise, information processing habits, and communication style across chats and calls.

To train the AI agent, users need to upload messages from work chats, documents, spreadsheets, emails, audio recordings, and screenshots. Once trained, the agent can write code based on technical specifications, answer questions, and — in a darkly humorous touch — even shift blame to others.

Why this matters

Colleague Skill didn't emerge in isolation. It arrived amid growing reports that employers in China were systematically requiring workers to document their workflows and decision-making processes in detail — only to lay them off once the documentation was complete. Management framed the exercise as "process optimization," but the collected data was actually fed into corporate AI training pipelines.

The backlash was swift. Users coined the term "worker distillation" to describe the practice — a reference to knowledge distillation in machine learning, where a large model's capabilities are compressed into a smaller one. The parallel struck a nerve across Chinese social media.

Fighting back: Anti distillation skill

In response, a counter-project called Anti distillation skill appeared online, aiming to help employees protect themselves from forced knowledge extraction. The tool rewrites archived documents to make them less useful for AI agent training.

«近日，github上一个名叫"同事.skill"的项目火了。4月3日，一博主表示，她开发了"反蒸馏skill"的项目。她表示，大家都是出来做牛马的，没人希望自己被做成skill，然后丢掉工作，所以自己发明了"反蒸馏skill"。希望大家在这个AI浪潮里都能活得久一点吧。» — 李老师不是你老师 (@whyyoutouzhele), original post

The creator of Anti distillation skill explained her motivation: nobody wants to be turned into a skill file and then lose their job. She expressed hope that people could "survive a little longer" amid the surging AI wave.

The "digitize everyone" trend

Beyond the core project, other users began publishing repositories with more eccentric use cases — offering to digitize an ex-girlfriend, a boss, or even oneself for handling business correspondence. While tongue-in-cheek, these forks reflect genuine demand for personalized AI agents that can mimic specific individuals.

The situation in China echoes global concerns. Surveys have previously shown deep anxiety among U.S. citizens about AI advancement and its potential impact on employment. Colleague Skill and its counter-tool illustrate the escalating tension between workplace automation and job preservation — a friction point that is unlikely to ease anytime soon.