Notes by Lex Toumbourou

Agents, in Agentic Reasoning, are entities powered by Large Language Modelss that can plan and take actions to execute goals.

They can be part of a Single-Agent Systems, or Multi-Agent Systems where multiple agents work together to achieve a goal.

Commonly, each agent is given a Persona, as well as access to tools (see Tool Use).

Some agents also contain a Memory component, where they can save and load information outside of their messages and prompts.