AI agent tooling provider LangChain raises $125M at $1.25B valuation – News

LangChain Inc., a startup that helps developers build artificial intelligence agents, has raised $125 million in funding at a $1.25 billion valuation.

Fortune reported today that IVP led the Series B investment. It was joined by Alphabet Inc.’s growth-stage CapitalG fund, ServiceNow Ventures, Workday Ventures, Cisco Investments, Datadog, Databricks and several others.

LangChain develops an open-source AI agent development tool of the same name. The software enables engineers to implement agents with as little as 10 lines of code. One of the ways LangChain speeds up development is by providing prepackaged building blocks that remove the need to create everything from scratch.

Another selling point of the tool is its unified application programming interface. OpenAI, Anthropic PBC and other AI providers distribute their language models through different APIs. As a result, changing an agent’s language model often requires switching it to a new API, which necessitates code changes. LangChain’s unified API makes it possible to switch AI models without code changes. 

Software teams that require more advanced features can use LangGraph, another open-source tool developed by LangChain. It facilitates the creation of AI agents that can run for extended periods of time and automatically recover from mistakes. LangGraph also makes it possible to implement human supervision features.

Companies working on even more complex projects can pair LangGraph with Deep Agents, an open-source tool LangChain released in July. The latter technology makes it possible to equip applications with reasoning features.

According to LangChain, Deep Agents includes a tool that enables AI agents to break down a complex task into multiple steps. The tool tracks the progress of each step and changes its processing plan if it encounters difficulties. A second Deep Agents component can create a dedicated sub-agent for each processing step to speed up output generation. 

Some tasks require AI agents to ingest a significant amount of new data. Under certain conditions, the volume of data that must be processed can exceed the capacity of an agent’s context window. Deep Agents includes a file system that increases the amount of information AI agents can use during processing. 

LangChain generates revenue with a paid product called LangSmith. It provides a code editor optimized for AI agent development. After engineers create a new agent, they can use LangSmith’s built-in testing features to determine whether it meets project requirements.

The tool also eases other development tasks. It includes a tool that makes it possible to deploy AI agents with one click. Once an agent is in production, the tool tracks metrics such as inference cost and latency. 

LangSmith’s observability features also monitor how users interact with AI applications. The tool can automatically identify user requests that an agent struggles to process. Developers can use that information to make improvements. 

According to News, LangChain generated between $12 million and $16 million in annualized recurring revenue as of June. A company spokesperson told Fortune today that its ARR has grown since then. LangChain is not yet profitable but claims to be spending its funding more efficiently than other high-growth venture-backed startups. 

Image: LangChain