Intermediate
Builder Path
For developers who know how to code but are new to building with AI. This path gets you from "I can use AI in the chat UI" to "I can build production AI apps" in around four hours.
Steps 12
Est. time ~3–4 hours
Prerequisites Programming experience
The Path
1
Model Families Overview
Know which model families exist (GPT, Claude, Gemini, Llama, Mistral, DeepSeek) and how they differ before you pick one.
15 min
2
Compare Models Side-by-Side
Interactive table: 48 models, filter by price, modality, open-weight. Use this as your reference when choosing a model.
10 min
3
Why Model Selection Matters
Frontier models are overused. This explains the real cost of defaulting to GPT-4o when a cheaper model would do.
10 min
4
Model Selection Cheat Sheet
Task-to-model mapping. Bookmark this — it's the fastest way to pick the right model for a specific job.
8 min
5
Cost Optimization Patterns
Prompt caching, batch APIs, routing, tiering — concrete patterns that cut costs without sacrificing quality.
12 min
6
What Is RAG?
Retrieval-augmented generation is the most important pattern in production AI. Understand the concept first.
12 min
7
The RAG Pipeline
End-to-end: document loading, chunking, embedding, retrieval, generation. Every step matters.
15 min
8
Vector Databases
Pinecone, Weaviate, pgvector, Chroma — how they work and when to use each. A key infrastructure choice.
12 min
9
What Is an AI Agent?
Agents are the next layer up from RAG. Learn the mental model before you build one.
12 min
10
Tool Calling
Tool use is how agents do things in the world — calling APIs, running code, searching the web. The core mechanic.
12 min
11
What Is MCP?
Model Context Protocol is the 2025 standard for connecting AI to tools and data sources. Critical to understand now.
12 min
12
Getting Started with Claude Code
Claude Code is the most capable AI coding tool available. This is your on-ramp to AI-assisted development.
15 min
After This Path
You now have the fundamentals of production AI development. Good follow-on topics:
- Model Economics section — go deeper on pricing, caching, and routing strategies
- Advanced RAG — hybrid search, reranking, query rewriting, evaluation
- Agent architectures — multi-agent patterns, frameworks (LangGraph, CrewAI, AutoGen)
- MCP deep dive — building your own MCP servers and connecting to existing ones
- Evaluation — how to measure whether your AI system is actually working
Checklist: Do You Understand This?
- Can you explain when to use GPT-4o vs a cheaper model like Haiku or Flash-Lite?
- Can you describe every stage of a RAG pipeline from document to answer?
- Do you know how vector databases work and which one to choose for a new project?
- Can you explain what tool calling is and how an agent uses it?
- Do you understand what MCP is and why it matters for 2025 AI apps?
- Have you set up Claude Code and used it on a real coding task?