Vercel's integration with the OpenAI Agents API divides a long-running AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → application into distinct responsibilities. OpenAI manages the agent's loop and session state; Vercel hosts the application and connects each session to a sandboxAn isolated environment where AI-generated code or agent actions run without being able to touch anything real.Full definition → for executing code and accessing files. The announcement describes signed webhooks processed through Vercel Queues to handle Sandbox creation and reconnection.
The workspace retains files across follow-up instructions, so later requests can continue work in the same session. Vercel describes this as an architecture that can scale to zero without an always-on worker, combining continuing workspace access with managed agent execution.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
sandbox — An isolated environment where AI-generated code or agent actions run without being able to touch anything real.
