Skip to main content

Photos & Slides

In-person photos from LangChain Interrupt 2026 (San Francisco, May 13–14, 2026), shot handheld from the balcony. They map to the session pages, which carry a synthesized summary plus the complete timestamped transcript for each recorded talk.

Note: These photos are from Day 1 (May 13). The Day 2 sessions are captured via transcripts on their session pages. Slide photos appear inline on the relevant talk pages as well as in the gallery below.


Day 1 — May 13, 2026

Venue

A packed house at LangChain Interrupt 2026, viewed from the balcony during the Day 1 keynote

The Interrupt 2026 crowd from the balcony

Keynote & Launches — Harrison Chase & Ankush Gola

➡️ Session page: The Deep Agents Era

Keynote slide: the deepagents-code CLI for coding agents with inference partners Fireworks AI, Baseten, and NVIDIA

Keynote slide: the agent harness from local to production (users, credentials, memory, auth, execution)

Keynote slide: Introducing LangSmith Context Hub

Keynote slide: Introducing LangSmith LLM Gateway (Beta)

Keynote slide: the deploy-a-deep-agent stack — Server, Model, Context Hub, Sandboxes, MCPs, UI

Keynote slide: the Build / Test / Deploy / Monitor lifecycle around Traces, with Govern underneath

Keynote slide: agent trace payloads grew over 6x in two years

Keynote slide: Introducing SmithDB, the database purpose-built for agent observability

Keynote slide: the SmithDB architecture — cluster manager, compaction/ingestion/query services, object storage

Keynote slide: Introducing LangSmith Engine, with customer quotes from Harmonic and campfire

Keynote slide: the LangSmith Platform — observability, evaluations, Context Hub, Gateway, Managed Deep Agents, Sandboxes over SmithDB

Introducing Managed Deep Agents — Sydney Runkle & Victor Moreira

➡️ Session page: Introducing Managed Deep Agents

Slide: Middleware enables business logic, policy enforcement, dynamic agent control, fault tolerance, context management, and toolsets

Slide: the Managed Deep Agents architecture — harness, runtime, context hub, sandboxes

How We Built It: LangSmith Engine — Ben Tannyhill & Vivek Trivedy

➡️ Session page: How We Built It: LangSmith Engine

Slide: the LangSmith Engine architecture — scheduled CRON trigger dispatching parallel per-issue fix runs

Slide: Rounding out the eval suite — diverse eval data, end-to-end evals, scenario-based tasks, sub-piece optimization

Building Frontier CX Agents — Carlos Pereira (Cisco)

➡️ Session page: Building Frontier CX Agents — Cisco

Cisco slide: the Renewals Agentic System Planner with slash-command shortcuts

Cisco slide: Lessons Learned — Renewals Agentic System

Cisco slide: Lessons Learned — Agentic Infra Automation

The Production System for Agents — Toyota

➡️ Session page: The Production System for Agents — Toyota

Toyota slide: Speed of Agent Development — from six months/six engineers to four days/one engineer

Toyota slide: the architecture-agnostic agentic platform (ToyotaGPT, TMMK, agent directory, LangGraph/LangSmith foundation)

Toyota slide: The Toyota Production System, quoting Taiichi Ohno

Toyota slide: Genchi Genbutsu — go and see, spot problems at the source


For official media, see the event page and LangChain's channels. Related: the LangChain Interrupt 2025 recap.