LangChain AI Updates: August 13, 2026

1. Harrison Chase Argues Managed Agents Are the Next Phase of Agent Building

LangChain. LangChain co-founder Harrison Chase traces the evolution of agent building from early frameworks toward what he calls managed agents, which bundle developer-supplied business logic, an agent harness, and production infrastructure into a single package running on managed infrastructure. The post argues that this model removes the burden of assembling runtime reliability, streaming, sandboxed code execution, context management, evaluation, memory, and authentication piece by piece, and relies on emerging standards such as AGENTS.md, MCP, and skill systems to control behavior. Chase cites Claude Managed Agents, Vercel’s Eve, and LangChain’s own Fleet and newly launched Managed Deep Agents as examples of the category. Source

2. LangSmith BYOC Reaches General Availability on AWS

LangChain. LangChain made LangSmith Bring Your Own Cloud generally available on AWS, letting enterprises run a managed LangSmith deployment inside their own AWS account and VPC while LangChain handles upgrades, scaling, monitoring, and backups. Sensitive data such as traces, datasets, and agent deployments stays in the customer environment, and because the deployment runs in the customer VPC, agents and sandboxes can reach internal databases, private APIs, and on-premises systems. The design uses a two-plane model, with a control plane in LangChain’s cloud communicating with the customer data plane over AWS PrivateLink, and is available to LangSmith Enterprise customers across 15 AWS regions. Source