LangChain: From Agent 0-to-1 to Agentic Engineering
Sequoia by Sonya Huang
LangChain just raised $125M and shipped 1.0. It’s now a big bet on agent engineering, not just a popular AI library.
Based on reporting by Sequoia, by Sonya Huang — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
LangChain just crossed a classic open-source threshold: a $125 million Series B and a full 1.0 release on the same day. Sequoia’s note frames the move as more than a funding milestone. It’s a statement that the team thinks the product, as rebuilt, is the right base layer for agent engineering.
The scale is already hard to ignore. LangChain says it gets 90 million monthly downloads and is used by 35% of the Fortune 500. The company has also expanded far beyond the original package. LangGraph handles low-level orchestration, memory, human-in-the-loop control and durable execution, while LangSmith covers observability, evaluation and deployment. That stack now reaches companies including Uber, Klarna, LinkedIn, J.P. Morgan, Clay, Cloudflare, Replit, Vanta, Rippling and Mercor.
The path here was not exactly linear. Harrison Chase started LangChain as a nights-and-weekends project in 2022, just as ChatGPT was taking off. The early pitch was simple: connect language models to tools and business logic, and they start looking like agents. That idea landed fast. By the time Sequoia led the Series A in early 2023, the project had more than 2,000 open-source contributors and was being used by more than 50,000 LLM applications.
But the source of the 1.0 release is not nostalgia; it’s a change in architecture. Sequoia says the old abstraction was a good starting point, but not low-level enough for the control developers wanted as the field matured. So langchain 1.0 is rewritten around LangGraph’s runtime, with pre-built agent patterns, improved model and tool integrations, and deeper customization. LangSmith has also become a bigger part of the story, with traffic growing 12x year over year.
There’s a nice bit of honesty in that shift. The company is not pretending one framework will stay enough forever; it keeps adding the plumbing underneath it. And now it has also launched an Insights Agent in LangSmith Observability and a no-code text-to-agent builder for business users. The message is clear: the team wants to own the boring, necessary middle of agent building, which is usually where real platforms are made.
My take — AI-written commentary, not fact-checked reporting
Open source wins when it turns into infrastructure, not just a popular GitHub repo with a nice logo. LangChain seems to understand that, and the move into LangGraph and LangSmith is the sensible one: less magic, more control, more tooling. The industry keeps rewarding anyone who can make agent building feel less like a demo and more like something an actual company can ship without praying.
Read more about this at: Sequoia
