LangChain just wrapped up their first-ever conference, Interrupt 2025, in San Francisco, and boy was it a hit! With more than 800 folks from all over the place showing up, it was quite the shindig. Big shots like Cisco, Uber, and LinkedIn spilled the beans on the latest and greatest in the AI agent world. The LangChain blog spilled some deets on how the event was all about showing off the potential of AI agents in different industries.
Keynote speakers went on about this newfangled thing called agent engineering. Apparently, it’s all the rage these days, kind of like software engineering but with a twist. They talked about the need for coding skills, understanding business workflows, and all that jazz to whip up some top-notch AI agents. Something about LLM (Large Language Model) applications needing a bunch of different models to work their magic. The LangChain package has been downloaded like crazy, beating out the OpenAI SDK in popularity. Go figure!
New products were also in the spotlight at the conference. The LangGraph Platform is now up for grabs, offering solutions for managing those long-running, stateful agents. You can deploy it on the Cloud, Hybrid, or do it yourself if that’s your jam. Then there’s the Open Agent Platform, a no-code agent builder for all you non-developers out there. LangGraph Studio v2 got a facelift, making it easier to visualize and debug agent interactions. And let’s not forget LangGraph Pre-Builts and LangSmith Observability, making agent-building a breeze.
Peeking into the future, LangChain teased a new technique called LLM-as-Judge, which sounds like a fancy way to test AI performance. They’re keeping it under wraps for now, but it involves human feedback to make sure everything runs smoothly. Interrupt 2025 was just the beginning for LangChain, as they plan to keep the party going every year. Innovation and collaboration are the name of the game, so keep your eyes peeled for more AI goodness from these folks.
