GitHub said at its Universe conference that it rebuilt the Copilot app for agent-led work. The app now covers specs, design talks and app telemetry. Copilot projects, available in the app now, run agent work across several repositories in managed parallel sessions. GitHub Enterprise customers can use every Copilot feature without a separate subscription and pay for the AI credits they use.
Copilot Fusion rates each task on reasoning, code generation, debugging and tool use, then picks one or more models to run it. InfoWorld reported that with Projects, a developer gives a goal plus context from repositories, mockups, Slack and meeting notes. Copilot spreads the work over parallel agents, tracks them and returns pull requests for review. GitHub said Fusion Max scored within two points of Opus 5.5 Max on the Artificial Analysis Coding Agent Index, at half the cost and 40% less time per task. That is GitHub's own claim. Not yet known: the price of an AI credit and a self-serve date.
If Fusion's cost edge holds on your own tasks and carries into credit pricing, which GitHub has not said, spend per task could fall. Test it on your workload. The argument here is that review becomes the limit, since GitHub itself said more code brings more pull requests to review. Jena told InfoWorld to restrict agent access to chosen repositories and require human approval before merges.