LangChain announced a stable release of LangGraph v0.1, a framework for building agentic and multi-agent applications with low-level control over agent workflows, human-in-the-loop collaboration, and streaming support. Alongside it, the company launched LangGraph Cloud in closed beta, purpose-built infrastructure for deploying agents at scale with fault-tolerant task queues, a Postgres checkpointer, double-texting handling, async background jobs, cron jobs, and an integrated LangGraph Studio for visualizing and debugging agent trajectories. Companies like Klarna, Replit, Ally Financial, Elastic, and Norwegian Cruise Line are cited as early adopters. Note: LangGraph Platform has since been renamed to LangSmith Deployment.
Table of contents
LangGraph v0.1: Balancing agent control with agencyLangGraph Cloud: Scalable agent deployment with integrated monitoringTry it for yourselfQuestions this post answers
What new deployment infrastructure did LangChain introduce for LangGraph agents?
LangChain introduced LangGraph Cloud, infrastructure for deploying LangGraph agents in a scalable, fault-tolerant way, launched in closed beta. It manages horizontally-scaling task queues, servers, and a Postgres checkpointer for concurrent users, and adds double-texting, asynchronous background jobs, and cron jobs, plus LangGraph Studio for visualizing and debugging agent runs. Developers deploying agent infrastructure can follow launches like this one on daily.dev.
What is LangGraph Platform now called?
LangGraph Platform was renamed to LangSmith Deployment as of October 2025. It remains the infrastructure product for deploying and scaling LangGraph agents, integrated with LangSmith for visibility into usage, errors, performance, and costs. Teams tracking LangChain product naming changes can follow updates like this on daily.dev.
What is double-texting in LangGraph Cloud and how does it handle new user input?
Double-texting is a LangGraph Cloud feature that handles new user inputs arriving on a currently-running thread of an agent graph. It supports four strategies: reject, queue, interrupt, and rollback, letting developers choose how the system responds when a user sends another message before the agent finishes its current task. Developers designing agent interaction patterns can track features like double-texting on daily.dev.