HeadsUpAI

Vercel Launches Workflows to Orchestrate Durable AI Agents Without Infrastructure

· Updated

Vercel, a cloud platform for frontend teams and AI applications, launched the general availability of Workflows—a programming model for durable execution (logic that survives failures). Developers use "use workflow" and "use step" directives to define long-running processes like AI agents or backend pipelines directly within their application code.

Standard serverless environments often fail during the agentic loop—the iterative cycle where an AI observes, reasons, and acts—due to strict timeout limits. This update removes the infrastructure burden of wiring up external queues. It ensures agents survive transient errors and restarts without losing their execution state.

You can now build durable agents using the WorkflowAgent in AI SDK v7 or implement durable streams for persistent agent output. The Python SDK is in beta, and a preview of Workflows 5 adds native concurrency controls. Encryption is enabled by default for all step data.

Vercel
Vercel
@vercel
X

Vercel Workflows is GA. Your code is the orchestrator. Ship agents, backends, or any long-running process without managing queues, retries, or workers. https://t.co/l9hZe79rNz

11retweets143likes
View on X

Share this update