Intermediate · 10 steps
Add AI to your app
A build-up path that takes you from "I've never called an LLM" to shipping an AI chat feature, then grounding it in your own data with RAG, then hardening it for production. You learn the handful of AI concepts that actually change how you build, in the order you hit them, and end with the cost/reliability controls that separate a demo from a live feature.
- 1Guide
Add your first AI feature
The on-ramp: orients you on what an AI feature actually is before you write a line of code.
- 2Term
LLM
You're about to call one on every request — know what the thing behind the API actually is.
- 3Term
Prompt
The prompt is your main control surface; everything downstream is prompt engineering whether you admit it or not.
- 4Term
Context window
The single constraint that most shapes chat and RAG design — learn it before you hit its limit the hard way.
- 5Recipe
Add an AI chat feature
Your first real feature: a streaming chat endpoint wired into the app, concepts made concrete.
- 6Term
Embedding
The mechanism that lets you search your own content by meaning — the foundation RAG is built on.
- 7Term
RAG
The pattern for grounding the model in your data so answers cite reality instead of hallucinating.
- 8Recipe
Build RAG over your own docs
Build RAG for real over your own documents, applying the embedding and retrieval concepts you just learned.
- 9Recipe
Add rate limiting
AI calls cost money per token — cap abuse and runaway usage before your first surprise bill.
- 10Recipe
Catch errors in production
LLM calls time out, get rate-limited, and return garbage; wire up error tracking so you see it when they do.