DevGPT

About

Dev-GPT treats a microservice request as a handoff across a virtual product manager, developer, and DevOps agent: the PM turns a description into a spec, the developer writes and debugs the code against tests, and the DevOps stage produces a Docker image with an optional hosted endpoint. When an implementation fails its checks, the loop retries alternative approaches rather than surfacing errors to the user. Generated services can include web-search capability via Google Custom Search, and a UI mode renders the result as a running app. It was one of the earliest end-to-end 'describe, deploy' demos from 2023; development stopped that same year, and users today would treat it as a reference implementation.