Book 11 — AI Agents¶
Autonomous cognitive loops, multi-agent state machines, and sandboxed execution environments — beyond calling an LLM in a loop.
Core Concepts¶
- Tool use & function calling
- Planning & Plan-and-Solve
- Memory & state
- Reasoning patterns (ReAct, Reflexion)
- Agent loops & POMDP formulation
Comprehensive Deep Dive:
Systems¶
- Multi-agent systems & topologies
- Tool orchestration (StateGraph & Pregel model)
- Retrieval / browser / coding agents (SWE-bench)
- Agent evaluation & trajectory benchmarks
- Agent security (prompt & tool injection, SSRF, sandboxing)
Comprehensive Deep Dive:
Mastery Roadmap
Each topic includes comprehensive mathematical formulations, architectural diagrams, runnable Python implementations from scratch, common debugging pitfalls, staff-level interview questions, and a 10-level mastery ladder.