Part V - Using Small AI for Agentic Applications

Agentic AI represents a fundamental shift from passive question-answering to active problem-solving. Rather than simply generating text in response to a prompt, an agentic system can plan a sequence of actions, call external tools, read and write files, execute commands, and iterate on its own output until a task is complete. The prevailing assumption has been that this level of autonomy requires massive frontier models with hundreds of billions of parameters. In this part, we challenge that assumption directly.

We explore two open-source tools that bring agentic workflows to the Small AI practitioner. First, we examine Hermes Agent from Nous Research, a self-hosted personal AI agent that lives on your own infrastructure, maintains persistent memory across sessions, and automatically builds reusable skills from its own experience. Because it is model-agnostic, you can point it at a local Ollama instance for routine tasks while reserving frontier models for complex reasoning. Second, we look at OpenCode, an open-source terminal-based coding agent with over 150,000 GitHub stars that supports 75+ model providers including local models. Its dual Plan/Build mode architecture gives you explicit control over when the agent is allowed to read versus write, making it especially well-suited for use with smaller models where you want to verify reasoning before granting write access.