An open-source company with 80,000 GitHub stars has been taken over by AI agents, with a supervisor Agent able to recruit Workers on its own, write code, assign tasks and track down bugs, according to a Nov 9 report by QbitAI, showing multi-agent collaboration at work in a real engineering setting. On the same day, TokenRhythm's Nov 5 release of the Agent-native model NeoHorse-1.5-Flash Preview was reported, with its Agent model RSI entering a second round of iteration.
AI agents take over open-source company as supervisor Agent recruits Workers
The open-source company, which has 80,000 stars, has handed its day-to-day engineering tasks to AI agents, QbitAI reported. The supervisor Agent not only writes code, assigns tasks and tracks down bugs, but also recruits new Workers on its own when it runs short.
When Workers run short, the supervisor Agent can recruit them itself!
The case is seen as multi-agent collaboration put into practice in a real engineering environment, marking a shift for AI agents from assistive tools to autonomous executors of engineering tasks.
TokenRhythm releases Agent-native model NeoHorse-1.5-Flash Preview
TokenRhythm released the Agent-native model NeoHorse-1.5-Flash Preview on Nov 5. Its RSI (iterative upgrade) has entered a second round of iteration, QbitAI reported, reflecting faster iteration of Agent models.
The release was reported on the same day as the AI agent takeover of the open-source company and is seen as a major signal that AI engineering is accelerating.
What to watch
The name of the open-source company, details of the takeover and the technical specifications of NeoHorse-1.5-Flash Preview have not been disclosed. The industry is watching whether this kind of multi-agent collaboration can be replicated across a wider range of open-source projects, and how iteration of Agent-native models will further drive AI engineering into practice.