Engineering the Agentic Stack
The Definitive Guide to Building AI Agents
A comprehensive engineering manual for mastering the architecture of intelligent systems—from LLM fundamentals to production-ready multi-agent orchestration.
Table of Contents
Part 1 — The Probabilistic Primitive
- 1
The LLM as a Kernel
FREETreating the language model as a new computing unit—understanding tokenization, context windows, and KV caching for system design.
15 min read - 2
Deterministic Constraints on Stochastic Outputs
FREETaming non-determinism in LLM outputs with JSON Schema and Structured Outputs to build reliable systems.
20 min read - 3
The Prompt Engineering Interface
FREETreating prompts as software: Architecture, versioning, Chain-of-Thought, and Few-Shot prompting patterns.
25 min read
Part 2 — The Agentic Architecture
- 4
Tool Use & Function Calling
FREEThe 'Hands' of the agent: Defining API schemas, handling hallucinations, and the execution loop.
20 min read - 5
Memory & Context Management
FREEStrategies for short-term conversation history and long-term knowledge retrieval (RAG).
20 min read - 6
Planning & Reasoning Loops
FREEMoving from simple chatbots to agentic planners: ReAct, Plan-and-Solve, and Self-Correction.
20 min read
Part 3 — Production & Scale
- 7
The Model Context Protocol (MCP)
FREEThe standard for connecting AI models to data—building MCP Servers for local files, databases, and internal APIs.
20 min read - 8
Multi-Agent Orchestration
FREEFrom single agents to teams: Router Patterns, Swarm Architectures, and Frameworks like LangGraph.
20 min read - 9
The Sandbox
FREEExecuting AI-generated code securely with Docker, Firecracker VMs, and E2B.
20 min read
Want full access to all chapters and series?
Get LM Frontier Pro