LangChain and NVIDIA announced the NemoClaw blueprint for building enterprise agent systems that achieve top-tier perfor

LangChain today launched the NemoClaw for LangChain Deep Agents blueprint, developed with NVIDIA to help enterprises build, evaluate, and deploy advanced open agent systems. The blueprint addresses a key concern for enterprises moving agents into production: the need to own their agent infrastructure and the intelligence embedded in agent memory, workflows, traces, model weights, and tuning data.

The blueprint combines three components. LangChain Deep Agents Code provides the agent framework and evaluation tools. NVIDIA Nemotron 3 Ultra serves as the underlying model. NVIDIA OpenShell runtime enables secure deployment and execution. Together, these components allow teams to customize agents for their specific workloads, deploy them with proper governance, and run them at substantially lower cost than proprietary alternatives.

The performance gains are significant. In LangChain's agent evaluation suite, NVIDIA Nemotron 3 Ultra with LangChain Deep Agents achieved an aggregate score of 0.86 at a cost of $4.48 per task. The next closest performing model cost $43.48, representing approximately a tenfold reduction in inference cost. LangChain achieved these results through deep customization of how the agent uses tools, manages context, and evaluates intermediate reasoning steps, tailored specifically to Nemotron 3 Ultra's strengths.

Harrison Chase, Co-founder and CEO of LangChain, emphasized that agent performance improves when teams tune the model and surrounding system together around their specific business patterns. Jensen Huang, Founder and CEO of NVIDIA, noted that enterprises can now build custom agents shaped by their proprietary data, workflows, and know-how, running safely wherever they operate.

The announcement is backed by a substantial partner ecosystem including EY, Baseten, Fireworks, Nebius, Crusoe, DeepInfra, and Together AI. EY is building an implementation practice around the stack for regulated industries, noting that open agent architectures provide the transparency and control that enterprise clients and regulators require. LangChain serves more than 7,000 customers, including NVIDIA, Bridgewater, LinkedIn, Workday, Harvey, and Rippling. The NemoClaw blueprint is available now for enterprises to evaluate.