EN
Back to the archive

The archive · Developer & Business Tools · Strategic decision · 2022–2026

LangChain's agent-engineering bet: open-source hit grows into a $1.25B unicorn

From a ChatGPT-era open-source library to an 'agent engineering' platform: LangChain raises $125M at $1.25B with 143k GitHub stars.

LangChain

The betThat turning LLMs into agents needs a new discipline, and a model-neutral open-source company can own the platform for building, testing and deploying reliable agents.Scaling

What the business is

Open-source framework plus commercial platform for building and running LLM agents: LangChain and LangGraph libraries, LangSmith for observability, evals and deployment, and a no-code Agent Builder.

Starting capital$10M seed led by Benchmark (Apr 2023); $25M Series A led by Sequoia at a $200M valuation (2024); $125M Series B led by IVP at $1.25B (Oct 20, 2025).

How it started

Harrison Chase, then an engineer at Robust Intelligence, started LangChain as an open-source project in late 2022, weeks after OpenAI released ChatGPT, to give LLMs access to real-time data and tools. It was an immediate hit; Chase co-founded the company with Ankush Gola, and it became one of the breakout projects of the generative AI era.

What happened

LangChain raised a $10M seed led by Benchmark in April 2023 and a $25M Series A led by Sequoia in 2024 at a $200M valuation. As OpenAI, Anthropic and Google built in capabilities that were once its differentiators, LangChain expanded from the library into LangGraph orchestration and LangSmith observability, evals and deployment. On October 20, 2025 it announced a $125M Series B at a $1.25B valuation led by IVP, alongside LangChain 1.0 and LangGraph 1.0 and a no-code Agent Builder.

How it ended up

Scaling: a $1.25B unicorn with LangChain+LangGraph at 90M monthly downloads, 35% of the Fortune 500 using its services, LangSmith trace volume up 12x year over year, and 143k GitHub stars — still independent and model-neutral.

Background

LangChain began in late 2022 as an open-source project by Harrison Chase, then an engineer at Robust Intelligence, weeks after OpenAI released ChatGPT. Its idea — 'chains' that connect large language models to web search, APIs and databases so they can act, not just generate text — was an instant hit, and Chase co-founded the company with Ankush Gola.

The bet was that turning LLMs into agents requires a whole new engineering discipline, and that a model-neutral open-source company could own that layer. LangChain raised a $10M seed led by Benchmark in April 2023 and a $25M Series A led by Sequoia in 2024 at a $200M valuation, growing into one of the most-starred projects in AI with 143,120 GitHub stars by August 2026.

As OpenAI, Anthropic and Google added built-in agent capabilities, LangChain expanded from the library into LangGraph orchestration and LangSmith observability, evals and deployment. On October 20, 2025 it announced a $125M Series B led by IVP at a $1.25B valuation, with LangChain and LangGraph 1.0 releases, a no-code Agent Builder, combined 90M monthly downloads, and 35% of the Fortune 500 using its services.

What has to be true

  • Timing: launching weeks after ChatGPT made LangChain the default answer to 'how do I connect an LLM to data?' at the exact moment everyone was learning.
  • Model neutrality kept LangChain useful as OpenAI, Anthropic and Google added native agent features; it works with every vendor.
  • The pivot from library to platform monetized the same workflow: LangSmith's observability, evals and deployment are where enterprises pay.
  • A 143k-star GitHub repository and 90M monthly downloads gave LangChain distribution no closed vendor could match.

What can be applied

Open-source adoption is distribution, not a business: LangChain turned a wildly popular library into revenue by selling observability and deployment for the workflow enterprises already used.

Aftermath

As of September 2026 LangChain remains independent and positions itself as 'the agent engineering platform': open-source LangChain and LangGraph 1.0 plus paid LangSmith for observability, evals and deployment, with a no-code Agent Builder in preview. It reports 90M monthly downloads across LangChain and LangGraph, 35% of the Fortune 500 as users, and 12x year-over-year growth in LangSmith trace volume, with customers including Cisco, Replit, Clay, Cloudflare, Workday and ServiceNow.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases