Pinned
Stop manually triaging agent failures.
Let LangSmith Engine fix it.
00:00
LangChain
@LangChain
Joined November 2022
- With LangSmith LLM Gateway, detection, investigation, and remediation happen on the same surface where agents are built. ✅ Fewer tools ✅ Fewer context switches ✅ Policy events that arrive next to the trace data that explains them Learn more: langchain.com/blog/introduci…
- At Interrupt, @AndrewYNg sat down with @hwchase17 for a fireside chat on the future of agents. Watch the session in full: youtu.be/OaRhpwz_TGM
00:00Want to see more sessions from Interrupt? Access our archive below: 
Read the guide to learn what teams are prioritizing across observability, evals, governance, security, and human-in-the loop reviews. info.langchain.com/guide/definiti…?- LangSmith lets your agents unpack. Understand what didn’t work out in production, identify what matters, and make continuous improvements.
00:00Get started with LangSmith today: - LangChain reposted
another banger from Sydney! i think this whole hierarchy of loops is still super early but some primitives we know work ex: verification as a primitive is so ridiculously important for non-slop semi-long-horizon work, it’s worth spending days to weeks making sure the - LangChain reposted
What are Online Evals? Most agent evals run "offline": a premade dataset of inputs goes through the agent, and an intermediate step or final output gets scored. They answer "is this version better than the last?" Online evals answer a different question: "is the agent still
- Agent governance isn't something that should be bolted onto agentic systems. LangSmith LLM Gateway lets you enforce the rules in the same platforms where agents are built, observed, and evaluated.
- LangSmith Sandboxes are the right layer when your agent needs to do something, not just say something. Here's when you want to reach for them 🧵 ✅ Your agent generates code and you want it to verify that code runs before responding ✅ You're building a coding assistant, CI✅ You're running multi-step workflows where state needs to persist across tool calls ✅ You need burst capacity (i.e. thousands of parallel environments for RL training or evals) that has to scale from zero in seconds ✅ You're accepting any user-supplied input that could end
