Vercel adds always-on workers to run more of your backend

Vercel, a platform for developers, has introduced Daemons, a new service for running always-on background workers, and made its Python support generally available at its Ship conference in San Francisco. Daemons, currently in public beta, allows developers to deploy containers and specify the number of instances to run in each region, ensuring they stay running and restarting them if they exit or crash.

The service supports pull-based workloads, where workers fetch jobs from a queue or database and process them as they arrive. Vercel CTO Malte Ubl explains that Daemons cater to sophisticated backend workloads that don't fit into request-driven work, such as processing tasks on a queue. For example, Vercel's own deployment pipeline uses Daemons, where containers package Python dependencies and media frameworks to generate playback files and thumbnails.

Daemons differ from Vercel's sandboxes in that they remain always-on but are not invoked by requests from the internet. Developers can deploy through Git or the Vercel CLI, and Vercel builds and stores the container image in its Container Registry. Meanwhile, Vercel has also made Drives, its persistent storage for AI agents, generally available.

Drives separates an agent's files from the sandbox executing its code, allowing agents to mount the drive in a new sandbox, change runtimes, and continue working with the files they already created. This service provides a hybrid of an EBS-style networked drive and object storage, with writes landing on a local NVMe disk and streaming asynchronously to object storage. Vercel encourages customers to switch from snapshots due to their large sizes and inefficiencies.