LangChain PM Victor Moreira on Managed Deep Agents
LangChain Product Manager Victor Moreira introduces Managed Deep Agents in public beta, covering the mda CLI workflow, identity management, and case studies from Madrigal and Consensus.
ImportanceLocalEvidenceE2 unreplicatedWrite-upQuick
At the Interrupt New York conference, LangChain Product Manager Victor Moreira introduced Managed Deep Agents, currently in public beta. This tool bundles the Deep Agents framework with managed LangSmith infrastructure, allowing developers to define agents as folders containing instructions, skills, and tools, then test and deploy them using the mda CLI.
Using LangChain's internal go-to-market agent as an example, Moreira demonstrated how the platform addresses key production challenges. These include caller identity across Slack and web surfaces, user and agent-scoped credentials that allow pausing for authentication, memory access control by channel, and versioned context within Context Hub.
The session also featured case studies on how Madrigal built pharmaceutical research agents and how Consensus deploys agents across its organization. Moreira concluded by highlighting new features in the 0.8 release, such as HTTP channels, user-scoped memory, and built-in search powered by Parallel.