Part 3 Generating Data with AI
Sometimes the data you need does not sit in a database waiting for a query. It is scattered across the public web, in product pages and listings that were never meant to be parsed. Part 3 is about going and getting it, and turning it into something structured and trustworthy.
Chapter 10 covers web scraping the old-fashioned way: fetching pages, reading HTML, and pulling out fields with hand-written selectors, so you understand what is actually happening before you hand any of it to a model. Chapter 11 brings AI into the loop, using a search API and an LLM to discover the right pages and extract structured fields without brittle, site-specific code. Chapter 12 ties the whole book together: you assemble the pieces into an agentic pipeline of small, single-purpose functions, discover, fetch, extract, and validate, with retries, logging, and a human-review queue for anything the model is not sure about.
This is also where the book's take on vibe coding comes home. You will see how much an AI agent can do across hundreds of messy pages, and exactly where a human still has to set the rules and check the output. The result is not a clever demo; it is a pattern you can repoint at almost any unstructured source.