LangChain's Second Birthday

LangChain is marking the second anniversary of the release of its Python package. In a post reflecting on the milestone, the company said that while it still feels early in building applications with LLMs, much has changed over two years.

LangChain began as a Python/JS open source library and has evolved into a company. Alongside the original langchain library, it now offers two additional products: LangGraph and LangSmith.

The company said its mission has remained consistent: to make it as easy as possible to build context-aware, agentic applications. The original tagline was to connect LLMs to external sources of computation and data.

According to the post, the ecosystem has stabilized, exploded, and matured. When langchain launched, there were approximately 3 LLM providers; today there are over 70 model providers and more than 700 different integrations in langchain. The focus has shifted from prototypes to production-ready applications.

LangSmith was developed starting in early 2023 to address observability and evaluation needs, with work alongside customers including Moody's, Elastic, Podium, and Rakuten. LangGraph was built after feedback that off-the-shelf chains were too rigid and preconfigured agents lacked reliability at scale; it provides a flexible orchestration framework for agents across a spectrum of autonomy, with human-in-the-loop support and moderation loops. Use cases cited include Replit's coding agent and Unify's sales agent.

The langchain open source package is now described as more stable, more comprehensive, and more production-ready. It reached its first stable release in early 2024, with only two releases with breaking changes since. The post also noted dedicated integration packages for top partners and standardized tests.

The company reported community growth over the past year: contributors doubled from 2,000 to 4,000; apps powered by LangChain quadrupled from 30,000 to 132,000; total downloads jumped from 20 million to more than 130 million across Python and JS. LangChain thanked community champions and ambassadors by name.

Looking ahead, LangChain said it believes the space is still in its early days and that there is a lot of tooling left to build, with langchain, LangGraph, and LangSmith as key pieces.