OpenAI has released an open-source project named Symphony – a SPEC.md specification built to orchestrate autonomous programming agents.
The system was designed by OpenAI engineers to eliminate a persistent bottleneck in their workflow – human attention exhaustion.
“Every engineer opened several Codex sessions, assigned tasks, reviewed results, managed the agent, and repeated the process. In practice, most people could manage three to five sessions simultaneously without issues before switching between them became painful.”
Developers quickly found it difficult to track which session did what, monitor stalled agents, and maintain a clear mental model.
Symphony solves this by organizing workflows around core project management elements like issues, tasks, tickets, and milestones instead of manual coding sessions.
“Symphony constantly monitors the task board and ensures that an agent is running for each active task to carry it through to completion. If an agent fails or freezes, Symphony restarts it. If new work appears, Symphony picks it up and begins organizing tasks.”
In this architecture, agent labor is decoupled from pull requests. An agent can analyze code, create implementation plans, and spawn task trees for other agents, while humans simply review them first.
This approach radically cuts down the costs of AI errors by shifting human responsibilities to reviewing and rejecting automated outputs.
Interestingly, Symphony is a SPEC.md file rather than a complex control suite. OpenAI created a reference implementation using Elixir for its excellent concurrency primitives.
Ultimately, OpenAI does not view Symphony as a standalone product but as an adaptable template for engineering teams.
