A 23-year-old AI entrepreneur, Yimao Zhou, is challenging the conventional wisdom that AI agents are making teams more productive by accelerating individual tasks. Instead, he argues that AI is exacerbating coordination problems and ignoring the real bottleneck: team collaboration. His company, Emagen AI, is building an operating system (OS) called Cagen to address this issue.

What's Wrong with Today's AI Agents

Zhou believes that most AI agent startups are optimizing the wrong unit of productivity. They focus on making individuals faster, but in doing so, they ignore the coordination costs that arise when multiple people work together. These costs include syncing progress, writing status updates, relaying information between people, and waiting for approvals.

According to Zhou, 60% of knowledge workers' time goes to these coordination costs, while AI tools only optimize the other 40%, which is the actual doing. This means that AI agents are accelerating individual productivity but making team collaboration worse. He argues that this is a directional error and that the industry should be building tools that address the real bottleneck: team coordination.

Background and Context

The idea of an OS-level agent, like Cagen, may seem radical, but it's not entirely new. Operating systems have been around for decades, and they've evolved to meet the needs of computing shifts. PCs had standalone software before Windows, mobile devices had individual apps before iOS and Android unified the experience, and cloud computing had scattered services before AWS became the infrastructure layer.

AI is following a similar path. We're currently in the "standalone tools" phase, where hundreds of agents are doing one thing well but not talking to each other. The platform phase is just starting, and the OS phase hasn't happened yet. Cagen aims to be that OS-level agent, an organizational operating system powered by AI.

Why It Matters to the Industry

The implications of Zhou's argument are significant for adult-industry platforms and operators. As teams become more distributed and global, coordination costs will only increase. If AI agents continue to optimize individual productivity without addressing team collaboration, it could lead to a breakdown in communication and decision-making.

Cagen's OS-level approach could help mitigate these issues by providing an orchestration layer that holds the organizational context, project history, quality standards, and dispatches work to the right people and agents at the right time. This could be particularly useful for adult-industry platforms that rely on complex workflows and team collaboration.

What Comes Next

Zhou predicts that most AI agent startups will be dead in three years because they're building at a layer that's about to get commoditized. The survivors will be companies that built at a layer the foundation models can't easily absorb, such as deep domain-specific data flywheels or OS-level agents like Cagen.

The real disruption won't be about which companies survive but what becomes possible. As teams become more fluid and atomized, an orchestration layer like Cagen will become essential for holding the organizational context and dispatching work to the right people and agents at the right time.

Key Facts

  • Zhou's company, Emagen AI, is building an OS-level agent called Cagen to address team coordination issues.
  • Cagen inverts the human-AI relationship by letting AI drive workflows and call on humans for judgment.
  • 60% of knowledge workers' time goes to coordination costs, while AI tools only optimize individual productivity.
  • Zhou predicts that most AI agent startups will be dead in three years because they're building at a layer that's about to get commoditized.
  • Cagen aims to provide an orchestration layer that holds the organizational context and dispatches work to the right people and agents at the right time.