AI BriefingAugust 2026: LangChain Newsletter — Managed Deep Agents, LLM Gateway, and More
Key point
LangChain has launched Managed Deep Agents and LLM Gateway in public beta.
Details
LangChain announced the public beta launch of Managed Deep Agents and LLM Gateway in its August 2026 newsletter.
Managed Deep Agents allows deploying deep agents to the LangSmith runtime with a single command, featuring built-in durable execution, sandboxes, and tracing. The LLM Gateway sits between agents and models to provide essential controls for production environments, such as cost management, rate limiting, model fallbacks, and sensitive data handling.
Additionally, the performance of the LangSmith Engine has improved, with agent issue detection capabilities enhanced by more than 2x and the performance of suggested fixes improved by 25%. Tuned Evaluators now allow automatically attaching quality feedback to production traces, and the BYOC (Bring Your Own Cloud) feature has officially launched on AWS, enabling users to run LangSmith within their own VPC.
On the open-source front, Deep Agents v0.7 was released, simplifying the default harness and reducing default input token usage by 65% while maintaining the same performance. Harrison Chase emphasized that general intelligence alone is insufficient when enterprises leverage AI as a core business or product, highlighting the importance of 'owning your intelligence'.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.