I introduce Agentic Design Patterns: A Hands‑On Guide to Building Intelligent Systems by framing intelligent agents as autonomous, goal‑oriented computational entities that perceive, decide, and act within an application’s infrastructure, which I call the “canvas.” I highlight large language models (LLMs) as the cognitive engines powering these agents and argue that structured design patterns are essential for managing state, tool use, memory, communication, and resilience.
- Key characteristics of agentic systems: autonomy, proactiveness, reactiveness, goal‑orientation, tool use, memory, communication.
The preface states the book extracts 21 key design patterns.
I link this overview to the broader context presented in the foreword, where the canvas metaphor is first introduced.
- Design challenges: persisting state across steps, deciding when/how to invoke tools, coordinating multiple agents, handling errors.
- Solution: a catalogue of 21 reusable design patterns that act as templates rather than strict rules, speeding development and improving reliability.
- Terminology decision: I treat the underlying infrastructure as a “canvas” to emphasize that agents operate inside a managed environment of state, tools, and communication channels.