Case Study: Capy scales parallel AI agent workflows with Daytona

Capy provisions 100k parallel sandboxes in one week with Daytona

Capy, an AI-native IDE, faced the challenge of scaling its platform to run thousands of parallel AI coding agents. Each agent required an isolated, stateful sandbox that could be provisioned quickly and run for extended periods. Building and maintaining this complex infrastructure in-house would have required significant engineering resources. They turned to the vendor Daytona for its sandbox runtime platform.

Daytona provided a solution with its lightweight SDK and declarative image builder, enabling Capy to provision sandboxes in under 90ms. The platform delivered secure, long-running, and stateful environments that could be paused and resumed. As a result, Capy provisioned 100,000 parallel sandboxes in a single week, saved three months of infrastructure engineering effort, and empowered its developers to run workflows lasting seven days. Daytona's infrastructure was crucial to Capy's scaling success.

Capy