Agent Knowledge vs. Tool Knowledge

Intro

This article explains the difference between Agent Knowledge and Tool Knowledge, and how they work together in Domo AI agents.

---

Agent Knowledge vs. Tool Knowledge

	Agent Knowledge	Tool Knowledge
What it is	Packaged expertise, training data, or indexed documents via Retrieval-Augmented Generation (RAG) or vector databases	Executable functions, APIs, or scripts (for example, a calculator, a web search, or an internal database query)
How it works	Ingested offline or built into the model; provides context, domain definitions, and behavioral guidelines	The agent dynamically formats parameters and triggers the tool, which performs an action and returns a definitive output
Primary purpose	Comprehend user intent, reason through problems, and synthesize information internally	Interact with the outside world, perform exact calculations, or retrieve real-time data
When to use	Abstract reasoning, understanding tone, and answering broad conceptual questions	Real-time information, executing multi-step tasks, and performing precise actions

Tip: Think of Agent Knowledge as the mind — it shapes how the agent understands and reasons. Tools are the hands — they reach out and act on the world.

Agent Tools vs. Agent Skills

Tools and Skills are related but distinct concepts:

• Tools — External interfaces (APIs, functions) that an agent invokes to perform an action. Use tools for deterministic actions, for example, “Send an email.”
• Skills — Internal cognitive patterns and workflows the agent possesses. Use skills for probabilistic reasoning, for example, “Draft a persuasive rebuttal.”