Agents are easy to demo locally. The hard part is shipping them inside a real app.
We published a deployment cookbook for @LangChain agents: full-stack examples with streaming UI, subagents, thread history, and production persistence notes across common JS frameworks 🚀
🧵👇
Users dismissed the LangChain_JS deployment cookbook for production AI agents as an unrequested tutorial because infrastructure problems remain unresolved.
No Digg Deeper questions have been answered for this story yet.
deployment cookbook for langchain agents!
Agents are easy to demo locally. The hard part is shipping them inside a real app.
We published a deployment cookbook for @LangChain agents: full-stack examples with streaming UI, subagents, thread history, and production persistence notes across common JS frameworks 🚀
🧵👇
Most JS teams ship from @nextjs first.
The cookbook example keeps the agent inside App Router: Route Handlers expose /api/threads/..., SSE streams live events, and @langchain/react connects the UI.
Useful when you want one deployable app, not a separate backend.
Not on @nextjs? The same deployment shape works across the JS ecosystem.
@nuxt_js: Nitro routes + @langchain/vue @sveltejs with SvelteKit: @Cloudflare Workers + Durable Objects for SSE replay
Or our recommendation: LangSmith Deployment via hosted LangGraph Agent Server + your UI
🧑💻 Code: https://github.com/langchain-ai/deployment-cookbook 📚 Docs: https://docs.langchain.com/langsmith/deploy-frameworks-and-platforms
The deploy targets (Next, Cloudflare, Deno…) are the easy part. The real "shipping" problem is statefulness — a thread that survives a refresh, a crash, a 3-hour async run. Once an agent has durable, resumable state, it stops being a demo and becomes infrastructure. That's the hard 90%.
@hwchase17 Clean abstractions matter more than ever for agent infra. Which runtime are you seeing the most stable deployments on?
@hwchase17 Another tutorial nobody asked for, production infra is still a mess.