What is langship.sh?

Langship is a self-hosted platform for deploying and managing AI agents, autonomous software programs that use language models to perform tasks. It is designed for teams who want to take an AI agent they have built and ship it reliably, with controls around safety, approvals, and governance baked in, rather than bolted on afterward. The core concept is a deployment pipeline modeled as a graph of steps. A pipeline might include stages like clone the code, build a container image, run a security scan, require a human approval, deploy to the target environment, and automatically roll back if something goes wrong. Each step is a reorderable node in the graph, so governance checks like policy enforcement or approval gates are visible first-class steps rather than hidden middleware. If a deployment crashes mid-run, it replays automatically from where it left off because each step is journaled. The platform is designed to deploy AI agents to different cloud runtime environments using the same pipeline definition, though today the deploy path that is fully built out targets AWS Bedrock AgentCore, with Kubernetes and Vertex AI support still in progress. It works with agents built using any framework, including LangGraph, LangChain, LlamaIndex, CrewAI, and AutoGen, so you are not locked into a specific AI development library. You connect it to a GitHub repository, configure environments, define your pipeline, and trigger runs either manually or automatically when code is pushed. Langship is entirely self-hosted, so your code, credentials, and run history stay on your own infrastructure. It comes with a web UI, a command-line tool, and real-time log streaming so you can watch runs as they execute. The backend is written in Go and licensed under Apache 2.0.