Deploy Self-Evolving Agents for Faster, More Secure Research with a Hermes Agent and NVIDIA NemoClaw

2026-06-02 · Source: NVIDIA Technical Blog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Cybersecurity & Data Privacy · Depth: Intermediate, medium

Summary

NVIDIA presents an open-source example for deploying a self-evolving AI agent using Hermes Agent and NVIDIA NemoClaw, secured by NVIDIA OpenShell. This solution facilitates product research by synthesizing data from internal sources like Outlook and Slack with public platforms such as GitHub and NVIDIA developer forums. The agent, powered by nvidia/nemotron-3-super-120b-a12b or a self-hosted Nemotron model on NVIDIA NIM, learns user preferences and patterns, writing new skills and memories that persist across deployments. NVIDIA OpenShell enforces a security-approved runtime, managing credentials and network access policies to prevent unauthorized data exfiltration. The article details installation steps, including cloning the nemoclaw-community repository and using openshell sandbox commands, and demonstrates teaching the agent a recurring report format directly through chat, which then persists and can be triggered from new conversations. Observability is provided via NeMo Relay and Arize Phoenix, recording traces in Agent Trajectory Format (ATIF).

Key takeaway

For MLOps Engineers deploying AI agents that handle sensitive data, NVIDIA NemoClaw with Hermes Agent and OpenShell offers a robust framework. You can securely integrate internal and public data sources. This ensures agent learning and skill persistence across deployments. Implement the provided open-source example to establish a secure, self-improving agent environment. Leverage network policies and sandboxing to mitigate data exfiltration risks. This approach streamlines agent management and enhances data security.

Key insights

Self-evolving AI agents can securely synthesize public and private data for research, learning new skills that persist across deployments.

Principles

• Enforce security with sandboxed runtimes and network policies.
• Agents should learn skills and persist state across deployments.
• Implement observability for agent decision-making and debugging.

Method

Deploy Hermes Agent via OpenShell sandbox by cloning the nemoclaw-community repository, configuring environment variables, starting host services, and using bring-up.sh. Teach skills through chat, then snapshot and restore agent state.

In practice

• Conduct product research across Slack, Outlook, and GitHub.
• Apply to sales research, customer support, and engineering triage.
• Use for competitive analysis and internal knowledge discovery.

Topics

• AI Agents
• Hermes Agent
• NVIDIA NemoClaw
• NVIDIA OpenShell
• Data Security
• Self-Evolving AI

Code references

• NVIDIA/nemoclaw-community
• nousresearch/hermes-agent
• NVIDIA/nemoclaw-community
• NVIDIA/NemoClaw
• NVIDIA/NeMo-Flow

Best for: AI Engineer, MLOps Engineer, AI Architect

Related on AIssential

• Nvidia’s version of OpenClaw could solve its biggest problem: security — Nvidia's NemoClaw offers an enterprise-grade, secure platform for deploying and managing AI agents based on the OpenClaw framework.
• Hermes Agent Masterclass — Hermes Agent offers a self-improving, persistent AI agent framework with unique learning, memory, and skill management capabilities.
• HERMES AGENT SETUP: the OpenClaw killer is here — Hermes Agent offers continuous self-improvement and skill generation for AI agents, emphasizing local, open-source model deployment.
• Hermes Agent Foundation Update: Real-Time Agents, DeepSeek V4 FREE, Native Windows Support, & More! — Hermes Agent's 0.14.0 update unifies AI tooling, boosts performance, and expands autonomous capabilities.
• Build enterprise-ready Agentic AI with DataRobot using NVIDIA Nemotron 3 Super — Transforming powerful LLMs into trusted, production-grade enterprise AI systems requires robust infrastructure beyond model capabilities.
• Run Local AI Agents with Faster Models and Multi-Node Clustering on NVIDIA DGX Spark — Integrated hardware and software solutions significantly accelerate local AI agent development and deployment, enhancing security and scalability.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by NVIDIA Technical Blog.