Mar 6
Loading video player...
In this video I walk through 3 less-known features that make agentic apps feel actually production-ready, using real code from the LangGraphJS repository. 📚 What you’ll learn - Render reasoning tokens (thinking / reasoning blocks) in the UI so users can follow how the agent approaches a problem, catch mistakes early, and build trust in results. - Reconnect to running streams after refresh/network blips so you don’t lose an in-flight agent run (better UX, fewer “it stopped working” moments). - Branch existing conversations + edit prompts mid-thread so you can correct course without restarting the entire chat (like “git branches” for conversations). If you’re building agentic products (not demos), these patterns are quick wins that materially improve trust, resilience, and iteration speed. 🔗 Repo / example app: https://github.com/langchain-ai/langgraphjs/tree/main/examples/ui-react 📖 Follow for more agent UI patterns: https://docs.langchain.com/oss/javascript/langchain/streaming/frontend