AI Engineer
← All organizations

AI agent frameworks and development infrastructure

LangChain

LangChain builds open-source frameworks and a commercial platform for teams developing AI agents that use data and take actions. Its namesake framework connects models and tools in customizable agent loops; LangGraph provides the runtime for custom workflows. Deep Agents packages context management, files, memory, and subagents for longer tasks. The commercial LangSmith platform lets developers trace behavior, evaluate results, deploy agents, and build them without code. It also works with agents built outside LangChain’s frameworks, giving teams a way to test and improve their systems without adopting the entire stack.

CEO Harrison Chase began LangChain as a Python side project in late 2022 and founded the company with Ankush Gola in early 2023. Its engineering approach emphasizes control over agent execution: LangGraph adapts the BSP/Pregel algorithm to support cyclic workflows and deterministic concurrency. Checkpoints save intermediate state so agents can recover from failures or pause for human approval and resume without restarting the task.

In March 2026, the company reported over one million practitioners using its open-source frameworks, more than one billion cumulative framework downloads, and over 300 enterprise customers for LangSmith. Its October 2025 financing raised $125 million at a $1.25 billion valuation, with IVP leading the round.

www.langchain.com ↗

4 talks

Newest first

3 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Harrison ChaseCo-founder and chief executive of LangChainLance MartinMember of Technical StaffVivek TrivedyHead of Applied Research

Start here

  1. Building Context-Aware Reasoning Applications with LangChain and LangSmith

    Start with Harrison Chase's architectural survey to understand how models, APIs, routing, and tools fit into a larger assistant, and where tracing and evaluation enter development.

    Harrison ChaseAI Engineer Summit 2023

  2. Architecting and Testing Controllable Agents

    Use Lance Martin's workshop for practical treatment of answer validation, agentic RAG state, document chunking, and retrieval precision-recall tradeoffs.

    Lance MartinAI Engineer World's Fair 2024

  3. Improving Agents is a Data Mining Problem

    Watch Vivek Trivedy's session to learn how production data can feed repeatable improvement loops, including supervised fine-tuning, reinforcement-learning approaches, and continual learning.

    Vivek TrivedyAI Engineer World's Fair 2026

  4. 3 ingredients for building reliable enterprise agents

    LangChain co-founder Harrison Chase explains how enterprise agents move beyond prototypes by targeting valuable, substantial work; combining agent autonomy with deterministic workflows through LangGraph; and making behavior observable and evaluable with LangSmith.

    Harrison ChaseAI Engineer World's Fair 2025

Messages from the stage

Choosing where agents control execution

Harrison Chase surveys architectures from chained calls to autonomous tool use, while Lance Martin contrasts fixed chains with model-directed control flow. Chase's enterprise session connects that architectural choice to deterministic workflows in LangGraph.

Making autonomy easier to accept

Harrison Chase discusses reversible actions, human correction, and interface patterns that reduce perceived risk when agents take on substantial enterprise work.

Turning execution records into improvements

Vivek Trivedy frames agent improvement as mining traces for failures, context-compaction effects, and differences in model behavior. His session also examines scaling that analysis with cheaper open models and using distillation and training approaches.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-27