Deploy LangGraph Starter

Deploy and Host LangGraph Starter on Railway

Production-wired LangGraph starter on FastAPI + Postgres 18: a demo graph that retrieves documents (pgvector when embeddings exist, full-text otherwise), drafts an answer, and stops for a human decision before anything is published — interrupt/resume over HTTP, with LangGraph state persisted in Postgres so pauses survive restarts. LLM drafting and LangSmith tracing are optional and env-gated: no keys, no crash, no fake data. Pre-deploy migrations, $PORT binding, and healthcheck included.

About Hosting LangGraph Starter

One-click deploy creates the FastAPI app service from the GitHub repo and a Postgres 18 service, wires DATABASE_URL to Postgres as a reference variable, and runs python scripts/migrate.py as a pre-deploy command so migrations run before every deploy. The app binds $PORT and Railway healthchecks GET /api/health. No API keys are required: without an OpenAI key the graph drafts from retrieved documents verbatim, and without LangSmith keys tracing stays off. Add OPENAI_API_KEY only if you want LLM drafting and embeddings, and LANGSMITH_API_KEY with LANGSMITH_TRACING=true only if you want tracing.

Common Use Cases

• Build a research assistant that retrieves real documents, drafts an answer, and waits for a human approve or reject before anything is published.
• Run durable interrupt/resume workflows where a paused run survives a server restart thanks to Postgres-persisted LangGraph state.
• Ship an AI agent starter with honest env-gated behaviour: retrieval falls back from pgvector to full-text search, and no LLM keys means no fake LLM output.

Dependencies for LangGraph Starter Hosting

• Postgres 18 (Railway Postgres) with the pgvector extension for embeddings and the LangGraph checkpointer
• Python 3.12, FastAPI, and LangGraph 1.2 (pinned in the repo)
• OpenAI API key (optional, only for LLM drafting and vector embeddings)
• LangSmith API key (optional, only for tracing)

Why Deploy LangGraph Starter on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying LangGraph Starter on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.