Skip to main content
Verglas is a serverless compute and data platform built around two modern cloud components:
  • Workers run stateless request and event handlers.
  • Durable Objects give a named object serialized execution, durable storage, SQL, alarms, and WebSockets.
The platform places and scales these components automatically. You write a Cloudflare-compatible Worker, declare the bindings it can use, and deploy it. There are no servers, shards, or worker pools to provision. Streams, pipelines, Iceberg tables, vector indexes, graphs, and query read models are built from the same Worker and Durable Object model. They are composable bindings—not separate runtimes you have to operate.

Build your first Worker

Install the CLI, log in, run locally, and deploy.

Understand the platform

See how Workers, Durable Objects, and bindings fit together.

Use built-in components

Add durable state, pipelines, Iceberg, vector search, graphs, and queries.

Start from a recipe

Build ingestion, lakehouse, deduplication, and semantic-search flows.

​
The programming model

An application can stay as small as one Worker, add a Durable Object when it needs coordination or state, and add data bindings when it needs ingestion or specialized storage. Each binding is explicit in wrangler.jsonc, so a Worker only receives the capabilities you declare.
Verglas follows the Cloudflare Workers authoring and configuration model. The runtime executes the deployed component on Verglas infrastructure.