How We Built LangSmith Engine | Interrupt 26
Until now, improving your agent has been a manual process of reading traces, looking for patterns, writing evals, and creating fixes. Now LangSmith Engine can run that cycle for you. It watches your production traces, clusters failures into named issues, diagnoses root causes against your code, and proposes fixes and eval coverage to keep regressions from coming back. You just review and merge improvements. At LangChain's agent conference Interrupt, Ben Tannyhill and Vivek Trivedy introduced LangSmith Engine and what it unlocks for teams running agents at scale. How We Built LangSmith Engine | Interrupt 26 00:00 Introduction and context 00:33 LangChain as the Agent Engineering Platform 00:50 Our go-to-market agent and the problems we hit 01:47 Why the current process is broken (customer pain) 02:48 What we set out to build 02:45 LangSmith Engine demo: the prioritized issue inbox 03:14 Engine proposes fixes and opens PRs 03:32 Custom online evaluators 03:46 Dataset examples for offline evals 04:28 Architecture overview: how Engine works end-to-end 05:18 Early customers: Clay, Vanta, Campfire 05:23 The first version: a wind-up toy 06:54 The false positive problem ("Show me the man") 07:53 Architecture deep dive: orchestration and sandboxes 09:49 Why traces are the most valuable input 10:47 Connecting source code for PR generation 11:10 Types of fixes Engine generates 12:02 Learning from customers: the preference problem 12:56 The agent overview: Engine's memory file 13:40 Passing to Viv: evaluating Engine itself 14:04 Why evals are the only answer 14:31 How we bootstrapped evals (dogfooding + synthetic data) 15:24 Building a diverse and rounded eval suite 16:14 How evals inform model selection and prompt decisions 17:41 Beyond evals: trusting user feedback 18:24 The self-improving loop: Engine improving Engine 19:04 Key learnings and closing summary 20:36 Thank you Extra resources: • Everything we shipped at Interrupt: https://www.langchain.com/blog/interr... • Meet LangSmith Engine: https://www.langchain.com/blog/introd... • About LangChain: https://www.langchain.com/

The Agent Development Lifecycle: Build, Test, Deploy, Monitor | Interrupt 26

Introducing Managed Deep Agents | Interrupt 26

Big Techday 26: Scaling LLM-RL for the age of agents - Konstantin Dunas, Prime Intellect

Software engineering at the tipping point

Andrej Karpathy: From Vibe Coding to Agentic Engineering w/ Stephanie Zhan

Beyond Components: Designing Generative UI for MCP Apps — Ruben Casas, Postman

The Future of AI Agents: What Will Interrupt 2027 Look Like? | Interrupt 26

Beyond the basics with Claude Code

How I deleted 95% of my agent skills and got better results — Nick Nisi, WorkOS

Harnesses in AI: A Deep Dive — Tejas Kumar, IBM

From Chatbot to Teammate: Cisco's Agentic AI Journey | Interrupt 26

How to build agents when the smartest AI isn't smart enough

Anthropic's Boris Cherny: Why Coding Is Solved, and What Comes Next

The Insane Genius of a Formula 1 Gearbox

Spec-driven Development: How AI Changed Everything (And Nothing) by Simon Martinelli @ Spring I/O 26

Should You Pick American, Chinese, Or European AI?

Conan O’Brien Delivers the Commencement Address | Harvard Commencement 2026

Model Context Protocol (MCP) Explained for Beginners: AI Flight Booking Demo!

Everything We Got Wrong About Research-Plan-Implement - Dexter Horthy

