Inside Managed Deep Agents: mda CLI, Context Hub, and Harbor Evals | Interrupt London
LangChain's Managed Deep Agents bundles harness and infrastructure to simplify production agent deployment
“models are only as powerful as the context that they're given”
LangChain PM Victor describes Managed Deep Agents, a product that tightly couples agent harness (execution environment, context management, delegation, human-in-the-loop) with production infrastructure including durable execution, task queues, MCP/A2A orchestration, and sandboxed code execution. The pitch is that teams can focus on business logic rather than infrastructure plumbing. A pharma company (Madrigal) is highlighted as an early customer using domain-specific research agents.