Editorial illustration for OpenAI releases Symphony, an open-source spec to let agents self‑manage tasks
OpenAI releases Symphony, an open-source spec to let...
Managing OpenAI's Codex AI was a lesson in human limits. Engineers found that manually assigning tasks to more than three to five concurrent agent sessions destroyed their own focus. The bottleneck wasn't processing power; it was the team's attention. Their fix was a fundamental reversal: they stopped giving orders and let the AI take its own tickets.
OpenAI has released Symphony, an open-source spec with a reference implementation that turns task trackers like Linear into a command center for Codex agents. Instead of developers juggling multiple sessions, the agents grab open tickets themselves, leaving humans to review the results. Some internal teams saw merged pull requests jump sixfold in the first three weeks, according to OpenAI.
With Symphony, each open ticket gets its own Codex agent and dedicated workspace that runs until the task is done, turning the task board into the place where work actually gets dispatched. Before Symphony, OpenAI developers ran several Codex sessions in parallel, handed out tasks, and chased down progress on each one, according to OpenAI. In practice, running more than three to five sessions at once was nearly impossible without constant context-switching tanking productivity.
"The agents were fast, but we had a system bottleneck: human attention," the developers write. That's what sparked the idea of flipping the workflow. Instead of supervising sessions, agents would pull their own work straight from the tracker.
A sixfold jump in merged pull requests for an internal team measures something specific: friction. The system’s goal was to clear the human from the machine's path. With Symphony, your role shifts from dispatcher to editor.
You review outcomes. The autonomous system fetches and grinds. By open-sourcing the spec, OpenAI asserts this attention bottleneck isn't theirs alone.
The bet is whether other developers will trust a ticket queue to agents that never, ever need a break.
Related Reading
Browse all OpenAI news →
Open Source D&B rebuilds 642 million‑business database after AI agents hit limits
Open Source OpenAI's GPT Transcribe Cuts Error Rate to 3.31%, Improving on GPT-4o
Open Source Study: 6.7% of Deepfake Requests on Hugging Face Targeted Children
Related Topic OpenAI researcher quits, citing distrust over ad‑driven engagement metrics
LLMs & Generative AI
Related Topic OpenAI launches GPT-Image 1.5 with precise editing for enterprise visuals
LLMs & Generative AI
From Archives Calculator Finds Prompt Caching and Lazy-Loading Save Tokens, Gains Small
From Archives OpenAI releases 1.5B‑parameter privacy filter with 50M params, 128‑expert MoE
From Archives Elon Musk and Sam Altman head to court in legal fight over OpenAI’s future
Business & Startups
From Archives ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI
Common Questions Answered
What problem did OpenAI's engineers encounter when managing Codex AI that led to creating Symphony?
OpenAI engineers discovered that manually assigning tasks to more than three to five concurrent agent sessions significantly destroyed their own focus and attention. The bottleneck wasn't processing power but rather the team's limited human capacity to manage multiple agents simultaneously, which prompted them to develop a solution that would allow AI agents to self-manage tasks instead.
How does Symphony change the role of developers from traditional task management?
With Symphony, developers shift from being dispatchers who assign tasks to being editors who review outcomes. The autonomous system handles fetching and processing work independently, removing the human from the machine's operational path and eliminating the need for constant human oversight and task assignment.
What specific improvement did the internal team measure after implementing Symphony?
The internal team measured a sixfold jump in merged pull requests after implementing Symphony, which demonstrates a significant reduction in friction within their workflow. This metric specifically shows how removing the human attention bottleneck allows agents to work more efficiently and autonomously without constant human intervention.
Why did OpenAI decide to open-source the Symphony specification?
OpenAI open-sourced Symphony because they recognized that the attention bottleneck limiting agent management isn't unique to their organization but affects developers broadly. By releasing the spec publicly, OpenAI is asserting that this is an industry-wide problem and testing whether other developers will trust autonomous agents with ticket queues and self-management responsibilities.
