Part 1: General Agents: Foundations
You've heard the headlines. AI will write all the code. The end of programming as we know it. Every developer needs to learn AI or get left behind.
It's easy to dismiss this as hype—another cycle of breathless predictions that fizzle into disappointment.
But 2026 is different.
Three independent trends are converging simultaneously: AI capability has reached production quality, mainstream adoption has passed the tipping point, and enterprises are betting billions on AI-native architecture. The evidence isn't coming from marketing teams—it's from academic competitions, industry-wide surveys, venture-backed startups, and billion-dollar acquisition decisions.
This part establishes the mental models and practical skills you need for AI-native development. By the end of Part 1, you'll understand not just that AI is changing software development, but how to think about this transformation strategically and how to work with General Agents effectively.
What You'll Learn in Part 1
Part 1 consists of six chapters that build from concepts to communication to tools to engineering discipline to specifications to principles:
Chapter 1: The AI Agent Factory Paradigm (Conceptual Foundation)
Foundational Concepts (Lessons 1-3): You'll examine concrete evidence proving 2025 is a genuine inflection point—ICPC perfect scores, 84% developer adoption, $3 trillion economy transformation. You'll understand the three core operational constraints of LLMs (statelessness, probabilistic outputs, context limits) that shape all AI-native development, and learn how your role evolves from coder to orchestrator through the OODA Loop.
Technical Architecture (Lessons 4-5): You'll learn the Five Powers that enable autonomous agents (See, Hear, Reason, Act, Remember), the three-layer AI development stack, and the AIFF standards ecosystem (MCP, AGENTS.md, Agent Skills) that makes Digital FTEs possible.
Business Strategy (Lessons 6-7): You'll discover how to position your domain expertise as a competitive moat, understand the Digital FTE business model and monetization strategies, and master the nine pillars of AI-Driven Development (AIDD).
Methodology & Synthesis (Lessons 8-9): You'll learn Spec-Driven Development—specifications as the new syntax—and how to orchestrate AI collaboration across the development lifecycle. Lesson 9 synthesizes everything into a coherent vision for building Digital FTEs.
Enterprise Sales (Lesson 10): You'll learn how to sell your Digital FTE solutions to enterprise customers in the $100-400 billion agentic AI market. You'll understand the four value propositions (Enabler, Implementer, Custom Developer, Workflow Disruptor), master consultative selling and outcome-based pricing, and learn to position yourself for the six factors enterprises prioritize when choosing AI service providers.
Chapter 2: Markdown - Writing Instructions
Before you can work with AI agents effectively, you need to write clear instructions. Markdown is the format used for CLAUDE.md files, project documentation, and agent instructions throughout this book.
Markdown Syntax (Lessons 1-5): You'll learn headings for document structure, lists for organizing information, code blocks for showing examples, and links and images for connecting to resources. Each element is taught through practical examples that mirror the files you'll write in subsequent chapters.
Chapter 3: General Agents - Claude Code & Cowork
Now that you understand why AI-driven development matters and how to write clear instructions, Chapter 3 teaches you how to work with Claude's General Agents in practice.
Claude Code Essentials (Lessons 1-8): You'll install and configure Claude Code, master the CLI interface, learn persistent project context with CLAUDE.md files, create custom instructions, build Agent Skills, and understand subagent orchestration.
MCP & Integration (Lessons 9-14): You'll integrate external systems via MCP servers, compile MCP to Skills for token efficiency, master the settings hierarchy, implement hooks for event-driven automation, and discover plugins.
Claude Cowork (Lessons 15-21): You'll transition from terminal to desktop with Claude Cowork, learn practical workflows, integrate with Chrome for web automation, use Connectors for Google Workspace/Notion/Slack, and work with document files (docx, xlsx, pptx, pdf).
Business Application (Lessons 22-26): You'll understand safety limitations and what's coming, master built-in document skills, learn the Code vs. Cowork decision framework, and discover how Skills become monetizable products.
Chapter 4: Effective Context Engineering with General Agents
You've learned the tools. Now learn WHY they work and WHEN to use each one. Chapter 4 introduces Context Engineering—the quality control discipline for Digital FTE manufacturing.
Foundation (Lessons 1-3): You'll understand why context quality determines agent value, learn the U-shaped attention curve and 70% threshold, and discover why Claude "forgets" things in the middle of long conversations (position sensitivity research).
Techniques (Lessons 4-7): You'll audit your CLAUDE.md for signal vs noise (the under-60-line rule), extract tacit knowledge into AI-consumable formats, master /clear vs /compact decisions, and design progress file architecture for multi-session work.
Advanced (Lessons 8-9): You'll prevent workflow drift with memory injection patterns and coordinate multi-agent systems using context isolation.
Integration (Lesson 10): You'll apply the full context engineering toolkit to build a production-quality specialized agent worth selling.
Chapter 5: Spec-Driven Development with Claude Code
Chapter 5 introduces Spec-Driven Development (SDD)—the methodology that transforms AI coding assistants from sophisticated autocomplete tools into production-grade development partners.
From Vibe Coding to Specifications: You'll understand why conversational "vibe coding" breaks down for production systems and how specifications as primary artifacts solve context loss, assumption drift, and architectural inconsistency.
The SDD Workflow: You'll learn the three levels of SDD implementation, from lightweight specs to full workflow orchestration with Claude Code's Memory, Subagents, and Tasks systems.
Practical Patterns: You'll master prompt patterns that reduce approval fatigue, front-load review at specification phase gates, and enable parallel execution of research and implementation tasks.
Chapter 6: The Seven Principles of General Agent Problem Solving
Chapter 6 synthesizes learnings from Chapters 1-5 into actionable problem-solving principles that generalize across all General Agent workflows.
The Principles (Lessons 1-7): You'll master Bash is the Key, Code as Universal Interface, Verification as Core Step, Small Reversible Decomposition, Persisting State in Files, Constraints and Safety, and Observability. Each principle is explored in both Claude Code (terminal) and Claude Cowork (desktop) contexts—showing how these are truly General Agent principles, not tool-specific tricks.
Integration (Lesson 8): You'll discover the Meta-Principle underlying all seven (general agents leverage computing fundamentals), see how principles combine in real-world workflows, and learn a decision framework for choosing between Code and Cowork based on task characteristics.
The Agent Factory Paradigm
The Agent Factory paradigm changes everything about how you think about building software:
- General Agents (like Claude Code) are thinking partners who help you explore problems and build solutions
- Custom Agents (built with SDKs) are specialized products deployed at scale
- Digital FTEs are AI Agents engineered to own entire business functions
The critical insight: General Agents BUILD Custom Agents. Claude Code isn't competing with the OpenAI SDK—it's using Claude Code to build the Custom Agents you'll deploy.
A Note on Approach
Every lesson includes "Try With AI" sections where you apply concepts to your own domain. This isn't abstract theory—it's immediately actionable.
By the end of this part, you'll be able to:
- Cite evidence for the 2024-2025 AI inflection point with concrete data
- Distinguish General Agents from Custom Agents and know when to use each
- Explain the Five Powers and how they combine for autonomous orchestration
- Navigate the AIFF standards (MCP, AGENTS.md, Skills) that enable Digital FTEs
- Use Claude Code and Cowork effectively for real-world problem-solving
- Apply the seven principles of General Agent problem solving
- Design a monetization strategy for your domain expertise
- Execute Spec-Driven Development workflows
- Sell your Digital FTE solutions to enterprise customers using consultative approaches
- Articulate your path forward through the rest of this book
The transformation of software development is underway. The question isn't whether AI will change how you build—it's whether you'll lead or follow.
Let's begin.