Vercel is expanding what developers can run on its platform beyond frontends. At its Ship conference in San Francisco on Thursday, the company unveiled Daemons, a new service for continuously operating background workers, and released Python support for general availability. The company also made Drives, its persistent storage designed for AI agents, generally available.
Daemons, now in public beta, allows developers to deploy a container and designate how many instances should operate in each region. The platform ensures those processes continue running and restarts them when they exit or crash. At launch, it handles pull-based workloads where workers retrieve jobs from a queue or database and process them as they arrive. The service sits at the opposite end of the spectrum from Vercel's temporary sandboxes—these workers stay on continuously but aren't triggered by internet requests, reaching out for work instead. Vercel says 5,000 teams have completed 150,000 container deployments since container support entered public beta. Python support is now generally available for frameworks including FastAPI, Flask, and Django, with the company reporting that 150,000 teams already operate Python backends on its platform, including engineers at OpenAI, Polymarket, and Exa.
Vercel CTO Malte Ubl tells The New Stack the company is pushing "further down the stack" to ensure more developers can use Vercel as their sole infrastructure provider. While the company's existing queues, durable workflows, and Services feature cover request-driven work, sophisticated backend workloads need different tools, he explains. "What really doesn't fit well into that bucket is what you do find in sophisticated backend workloads," where developers want to process jobs from a queue with precise control over concurrency. Vercel itself runs a substantial portion of its control plane API on the platform—about a billion requests monthly, according to Ubl—and is now migrating many asynchronous workloads onto Daemons.
The company positions Drives as addressing how agents change storage requirements. The service separates an agent's files from the sandbox running its code, allowing an agent to mount the drive in a new sandbox, switch runtimes, and keep working with files it already created. Ubl describes it as a hybrid of an EBS-style networked drive and object storage: writes land on local NVMe disk and stream asynchronously to object storage, while reads come from local disk when data is already present or stream in block by block. Drives supports read-only mounting for multiple agents to share datasets and forking for testing different approaches in independent writable workspaces. The service is available on all Vercel plans with default capacities of 1 TiB, or 1 GiB on Hobby tier, and configurations reaching 16 TiB. For teams already hosting frontends on Vercel, each backend service they can keep on the platform means one fewer deployment system to maintain. The shift reflects how platform providers are racing to capture more of the development stack, turning convenience into strategic lock-in.