Back to feed
Dev.to
Dev.to
7/24/2026
How to run Hermes, a self-improving personal AI agent, fully local with QVAC

How to run Hermes, a self-improving personal AI agent, fully local with QVAC

Short summary

Hermes Agent from Nous Research is a personal AI agent with persistent memory, customizable persona (SOUL.md), and self-improving skills that runs fully locally using QVAC, an open-source SDK from Tether. Running locally keeps persona data, memory, and learned skills on-device, works offline, eliminates per-token costs, and prevents silent model swaps. The guide recommends Qwen3.6-35B-A3B as a capable MoE model for tool-heavy agent workloads.

  • Hermes Agent features persistent memory, persona customization via SOUL.md, and background skill curation
  • QVAC provides local model inference across NVIDIA, AMD, Intel, and Apple Silicon with an OpenAI-compatible server
  • Local deployment ensures privacy, offline capability, zero per-token cost, and no silent model swaps

Generated with AI, which can make mistakes.

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more