Pick an AI product, or describe your own, and see the building blocks it likely runs on, each one explained and linked. A map from the products you use to the concepts behind them.
An educational illustration of the building blocks a product like this typically uses, inferred from how it behaves, not confidential architecture.
Every AI product is a stack of a few recurring building blocks. The same handful, a context window, retrieval over embeddings in a vector database, tool calling, structured outputs, MCP, and memory, show up again and again. Recognise them in the tools you already use and the whole field gets less mysterious.
For the preset products, the mapping is a reasonable read of public behaviour, not inside knowledge. For your own description, it’s a keyword match meant to point you at the right concepts to learn, not a precise spec. Everything runs in your browser.
It’s an educated, public-behaviour read of the building blocks they almost certainly use, not leaked internals. The point is to connect familiar products to the concepts that power them.
It matches keywords in your description to the relevant concepts, an honest heuristic to point you at what to learn, not an AI analysis.
Start with AI Explained, or jump to RAG, embeddings, and tool calling.