LangChain Inc., a startup with a set of popular open-source tools for building artificial intelligence agents, has raised $300 million in funding.

The Wall Street Journal reported today that Alphabet Inc.’s CapitalG fund led the round. It was joined by IVP, Sapphire Ventures, GIC and DFJ Growth. LangChain is now worth $3 billion.

San Francisco-based LangChain is best known for the eponymous open-source framework it released in 2022. The software reduces the amount of work involved in building AI agents. LangChain makes it relatively simple to connect an AI agent to a large language model, integrate it with tools and add a system prompt. The framework also enables developers to customize details such as the format of prompt responses.

LangChain’s second open-source tool, LangGraph, is designed for situations where a task is too complicated to be automated by a single agent. LangGraph can link together multiple LangChain-powered agents into a single automation workflow. It can also add in other elements such as a mechanism that requires humans to approve sensitive actions.

The third component of LangChain’s open-source portfolio is Deep Agents. It extends LangChain and  LangGraph with features that make it easier for agents to perform complex work. There’s a virtual file system in which agents can keep data, a work planning tool and a subagent generation feature.

LangChain generates revenue with a suite of paid products called LangSmith. Developers can use it to not only build agents but also to maintain them once they’re in production.

An AI agent usually goes through multiple iterations before it’s released to users. LangSmith enables software teams to save those iterations in a centralized repository. If an update introduces a bug into an agent’s code base, developers can use LangSmith to quickly restore an earlier, bug-free version. Furthermore, they can compare the output quality of different agent versions to identify the one that should be deployed to production.

LangSmith also includes a set of observability features. It generates traces, logs that describe every single step of the workflow through an agent processed a prompt. LangSmith analyzes those traces to find technical issues. From there, it prioritizes them based on their severity and generates code fixes.

After its previous funding round last year, LangChain disclosed that its open-source tools were being downloaded about 90 million times per month. Its LangSmith suite, in turn, is used by Autodesk Inc., Cisco Systems Inc. and a long list of other tech firms.

LangChain reportedly plans to nearly double its headcount to 850 employees by the end of 2027. The company will hire infrastructure experts, AI engineers, salespeople and marketers. 

Image: LangChain