Welcome to the AI Barfed Series! This is our second post in the series. Today we’re showing how to develop a web scraper using our favorite everyday AI and Apify.
More often than not, we don’t know what we’re doing. Jordan figured his web scraper build would eat up the entire livestream. Instead, he finished in 37 minutes and we spent the rest of the session doing what we do best: two guys talking about random stuff while ~100 GTM engineers wondered why they were still watching.
The scraper worked, though. By the time Jordan was done ping-ponging through ChatGPT, Claude, Claude Code and Gemini while copying cURL commands he couldn’t interpret, we had a functioning Apify scraper that could pull dealer data faster than any of us expected. More importantly, we’d accidentally created a framework that Cannonballers can follow the next time they needed data that wasn’t conveniently packaged in an API.
The “Treat Me Like an Idiot” Framework
Jordan admitted, out of the gate that he might stumble around a bit. It turned out to be the most valuable part of his approach. Instead of pretending to understand web development, he leaned into his ignorance and let everyday AI do the heavy lifting.
Here’s how it actually worked:
Start with voice, not typing. Jordan used Super Whisper to narrate what he was seeing on the Generac Dealer site. “I’m typing 95125, I’m clicking search, I only care about home generators.” This captured context without getting bogged down in technical details he didn’t understand.
Feed everything to the AI first. He copied the entire HTML source of the search form (thousands of lines of code he couldn’t interpret) and pasted it into Gemini. The million-token context window meant he could dump everything and let the model figure out what mattered.
Let the models ask you questions. Instead of trying to direct the technical implementation, Jordan told Gemini his goal and waited for questions. The model asked about infinite scroll, email extraction, geographic limitations; things Jordan might have missed if he’d tried to plan the scraper himself.
Screenshot the network tab, don’t interpret it. When Jordan needed to find the backend API, he took screenshots of the browser’s developer tools and asked the AI to identify which requests mattered. He didn’t try to understand what he was looking at.
This approach works because it forces you to document your requirements clearly while letting the AI handle technical decisions you’re not qualified to make.
Why Apify
The backend discovery was a breakthrough moment. Jordan found that cURL command calling Generac’s dealer API and realized something important: building a custom scraper from scratch would take weeks, but Apify could turn that API call into a production scraper in minutes.
Apify’s $50/month platform handles the infrastructure that usually kills scraping projects. Anti-bot detection, IP rotation, scaling, data storage — all the technical complications that turn a simple data-gathering task into a development nightmare.
