LangChain provides open source frameworks and LangSmith, an engineering platform, that help developers build, test, and deploy reliable AI agents in production. Trace agent behavior with native observability, evaluate performance using real-world data and automated scoring, and scale deployments with fault-tolerant infrastructure supporting long-running, human-in-the-loop interactions. The platform powers teams across startups and enterprises, including 5 of the Fortune 10.
LangChain provides open source frameworks and LangSmith, a comprehensive engineering platform, that enable developers to build, test, and deploy reliable AI agents in production environments. The platform combines native observability through structured tracing—breaking down agent runs into step-by-step timelines with analytics and AI-driven insights—with evaluation capabilities that capture production traces, convert them into test cases, and score agent performance using a mix of human review and automated scoring. LangSmith's deployment infrastructure handles the unique demands of long-running agents, including async collaboration with humans and other agents, with built-in memory, conversational threading, durable checkpointing, and fault-tolerant scaling. The platform supports multiple frameworks and SDKs (Python, TypeScript, Go, Java) and integrates with popular agent frameworks and OpenTelemetry, allowing teams to instrument any agent stack. LangSmith Fleet extends the platform with autonomous agents for routine tasks like research and follow-ups, improving through user feedback and enterprise security controls. Trusted by 5 of the Fortune 10 and 6,000+ active customers, LangChain powers teams from startups to global enterprises seeking to ship agents that measurably improve across the development lifecycle.
LangChain explicitly states it will not use Customer Data to train on, develop, or otherwise improve its products. Secti
No training on user data means no opt-out is necessary.
Moderate risk. Suitable for internal data but avoid confidential information.
LangChain explicitly states it will not use Customer Data to train on, develop, or otherwise improve its products. Section 4.2 provides clear protection against AI training on user content.
“LangChain agrees that it will not use Customer Data to train on, develop, or otherwise improve its products.”
Customer grants LangChain a 'limited-term license' to use Customer Data only to provide the platform. However, Feedback (Section 4.4) grants LangChain an 'irrevocable, perpetual, worldwide, royalty-free license' without restrictions, which is broad for that category.
“Customer hereby grants LangChain a limited-term license to utilize Customer Data as necessary for LangChain to provide the LangSmith Platform in accordance with this Agreement.”
Third Party Products may receive Customer Data if Customer enables them (Section 2.6). Operational Metadata can be shared if aggregated/anonymized (Section 4.3). Service providers receive data for operational purposes (Privacy Policy). Sharing is somewhat controlled but occurs.
“If Customer enables a Third Party Product, Customer acknowledges that the LangSmith Platform may transmit or exchange Customer Data with the Third Party Product as authorized by Customer.”
Customer can request data export within 30 days post-termination and deletion within 30 days of written request. However, the 30-day window is relatively short and data in routine backups may be retained longer per Section 5.2.
“Within thirty (30) days of Customer's written request following termination or expiration of this Agreement, LangChain shall delete all Customer Data in its custody and control.”
Customer retains ownership of Customer Data and Confidential Information (Section 4.2). LangChain retains ownership of the platform and all derivatives/modifications. Clear separation of IP ownership favors customer protection.
“As between Customer and LangChain, Customer Data and Customer Confidential Information are and will remain owned exclusively by Customer, the User, or their licensors, as applicable.”
The ToS references a Data Processing Addendum (DPA) for GDPR compliance and mentions data protection laws, but does not explicitly list SOC2, HIPAA, or ISO certifications in the provided text. Privacy Policy mentions UK/EEA compliance but lacks comprehensive certification details.
“Where Customer's use of the LangSmith Platform for Cloud or Hybrid Deployments includes the processing of Personal Data subject to the applicable data protection laws, it will be governed by the Data Processing Addendum located at langchain.com/DPA.”
Track and measure your AI tools with Olakai's observability platform.
Explore Olakai