

Building Better Agents: The Agent Improvement Loop (New York)
Join us for an evening in New York with Palash Shah from the Applied AI team at LangChain.
Most teams can build agents. Far fewer have a system to improve them.
In this meet-up, Palash will walk through the agent improvement loop and how teams use traces as the foundation for continuous improvement. Every trace is useful data: a record of real behavior that can be enriched with evaluators, human feedback, and annotations to help teams understand performance, identify issues, and improve faster.
We’ll cover how to:
Capture traces of real agent behavior
Enrich traces with evaluations and feedback
Turn production behavior into reusable datasets
Use those insights to iteratively improve systems
We’ll also discuss how teams operationalize this loop in practice, from understanding agent behavior to building evaluation workflows and repeatable improvement systems.
Featuring:
Palash Shah, Applied AI Engineer @ LangChain
Please note this event is fully in-person in New York and will not be live-streamed online.
📍 Location: Flatiron – address provided upon approval.
Agenda ⏰
6:00 PM: Welcome + Food/Drinks
6:30 PM: Presentation with Palash Shah (Applied AI Engineer @ LangChain) - Building Better Agents: The Agent Improvement Loop
6:50 PM: Q&A Session
7:00 PM: Networking + Food/Drinks
8:30 PM: Event Ends
👀 To avoid disappointment at the door, we ask that only those with a confirmed QR code/email attend. If you didn't make it this time, we hope to see you at the next one!
About LangChain:
LangChain provides the agent engineering platform used by millions of developers globally. LangSmith brings observability, evaluation, and deployment together in one platform so teams can iterate and improve production agents. LangSmith works with any stack, including our open source frameworks like Deep Agents and custom code.
LangSmith powers top AI teams, from startups to global enterprises -- including Cloudflare, Toyota, Rippling, and Clay. Learn more: https://www.langchain.com/