AI Agent Architecture — ReAct, Memory, Planning and Multi-Agent Systems · Tool-Augmented LLM — Tools, Functions and the Environment

Why an LLM Needs Tools — the Knowledge and Capability Gap

Tool-Augmented LLM — Tools, Functions and the Environment

Introduction

Large language models generate text from weights learned during training — they are static artifacts with a knowledge cutoff, no access to external data, no way to perform computations with guaranteed precision, and no state between calls. This lesson analyses the four fundamental limitations of a "bare" LLM (knowledge cutoff, no code execution, no persistence, no side effects) and builds the architectural rationale for the tool-augmented LLM pattern.