As OpenAI and Anthropic expand their tools for building AI agents, LangChain co-founder and CEO Harrison Chase has a message for developers: Own your intelligence.
This means controlling the models, context, and systems surrounding the AI agent rather than relying entirely on a single model provider.
Chase made this argument at the LangChain Interrupt conference in London on October 13, nearly a year after the company raised $125 million in funding at a $1.25 billion valuation. This financing helps Langlian become a major player in the developer infrastructure field. Now Chase must prove that its software will still be useful even as the largest artificial intelligence companies develop competing products.
“Langlian’s mission is to empower every company to own its own intelligence,” Chase said in his keynote speech.
For businesses, this means being able to build agency around their own information and expertise while retaining the flexibility to change the underlying model.
Why Chase wants developers to control the wiring harness
Chase’s argument focuses on what developers call agent tools: systems that control how AI models use tools, access information, and perform tasks.
An enterprise might use a powerful model provided by OpenAI or Anthropic, but the company’s own data, business processes, and specialized tools determine how useful the model is in practice.
Chase believes that organizations should retain control of these components rather than handing them over to the model provider.
He is particularly interested in Model optionalitywhich allows developers to choose between models based on performance, cost, or suitability for a specific task.
This does not necessarily require using an open weight model. Chase also advocates keeping the harness open and controllable, explaining that open source software can provide developers with greater flexibility.
This difference has commercial significance for Langlian. If companies want to be able to switch model vendors without rebuilding their applications, independent software platforms can help them do this.
Langlian expands its enterprise tools
Following Chase’s keynote speech, he conducted a review of LangChain’s latest releases on October 15th, covering products launched in previous weeks.
These include LangSmith AI Search, which allows developers to search for records of agent activity using natural language questions. Instead of just relying on traditional filters, they can look for conversations where customers are frustrated or where agents expose sensitive information.
LangSmith Engine v2 adds automated testing and tooling designed to detect issues before agents are deployed. The update, first announced in September, also identifies inefficiencies and tests recommended fixes.
Managed Deep Agents v0.9 was released on October 7, introducing scheduled tasks, per-run configuration, and Slack reactions. Meanwhile, LangSmith Fine-Tuning allows developers to use records of agent activity to train and evaluate models for specialized applications.
Together, these products expand Langlian’s business beyond the original open source framework that helped popularize the company. Once companies start using agents in day-to-day operations, they solve problems that become more important: monitoring performance, identifying errors, and improving reliability.
They have also expanded the services LangChain can provide to paying enterprise customers.
The market is increasingly crowded
LangChain started as an open source project in October 2022, and Chase and co-founder Ankush Gola established the company the following year.
Its products now include LangChain and LangGraph, which help developers build and coordinate agents, and LangSmith, a business platform for evaluating, deploying and monitoring agents.
The company says it works with 35% of Fortune 500 companies and has accumulated more than 1 billion open source downloads. These numbers are reported by companies and do not determine how much revenue individual customers generate.
Chase’s vision also reflects broader concerns among technology executives about who controls corporate artificial intelligence systems.
Microsoft CEO Satya Nadella emphasized the importance of organizations having a learning process to improve artificial intelligence applications. Palantir CEO Alex Karp also made a similar point, saying that enterprise customers want to control their computing infrastructure, models and data.
At Interrupt London, attendees can experiment with agent behavior in the LangChain Agent Arena, where they train agents to play Tetris and compete on leaderboards.
Outside the arena, the business interests are considerable. OpenAI, Anthropic and other model developers are expanding the tools they provide for building agents, potentially reducing the need for external software.
Chase’s answer is to let LangChain work no matter which model the customer chooses. Its latest product shows how he plans to do that: by helping companies manage the systems they build around these models.
