One Workflow to Scrape Anything: n8n + Zyte AI Tutorial

Stop me if this sounds familiar. You open the dev tools and for that one specific div class, write your selector and it works perfectly for about 3 days. Then the site updates, the layout shifts and your entire workflow breaks. It's the maintenance step and honestly it's the worst part of web scraping. But what if we could flip the script instead of trying to fix our selectors? What if we just got rid of them entirely? Today I'm going to show you an automated AI scraper that does exactly that. Whether you are scraping an online store, a job board or an article site, this workflow adapts to that layout automatically without you writing a single CSS selector. If you are running this using any and the site API, it's basically one workflow to scrape anything. Let me show you how it works. Now, before we start, here is the best part. You don't need to build this complex logic from scratch. We have packaged the entire workflow into a template that you can grab right now for free. The link is in the description. Just follow the link. Go to the edit and template page. Hit use for free. Copy the template and you are halfway there. I'm running an locally using docker because I like keeping things on my machine. But this works exactly the same way on any cloud. As long as your instance is running, you are good. Now to power this workflow, we use the zite API. This will allow us to bypass manual selectors completely. You will just need to grab a free API key from the dashboard to get started. Okay, back to any paste the workflow we just copied from the template page and your scraper is ready to work. The best part about this workflow is that it comes with a built-in control center. So, you don't have to mess with the nodes directly. Let's start simple and execute this. I want the details for this specific book. So, I just copy and paste the URL. Select online store as my category. paste in the API key and hit submit. Next, it will ask me for the extraction goal and for that I will pick scrape details of a single item. Hit submit again. Now, if you follow the path and it is sending this to Zite, but here is the magic. I didn't tell it where the price is. I didn't tell it where the title is. All because it uses Zit's AI powered extraction. And the workflow is completed. Let's see the output. We got title, price, availability, image, URL, all structured, all clean, zero selectors required. Now you might be wondering about all these pipelines. It all starts here at the smart routing stage. This switch node acts as traffic control. The first three lanes are for the AI extraction pipeline. If you select a product, a job post or an article, you go here. The fourth lane is dedicated to the SER pipeline for search engine results. And the fifth lane is manual mode for when you just want row HTML. Now inside that top AI pipeline, we have this config generator. This is the brain of the operation. It takes your row form input like online store and converts it into the specific technical keys the Z API understands like product. It also checks your goal to decide if we need to turn on the crawler or just scrape one page. It handles all that complexity for you. Okay, getting one item is easy. But what if I want all of them? Let's take this category page. It has multiple pages of books and I definitely don't want to click next manually. So I will copy and paste the URL here. Select the category and this time specifically I am selecting scrap details of all items on all pages. As you can see this triggers the fifth path in our workflow the mass scroller. It has to walk through every page to find the links. This takes a moment. Okay, phase one is done. We have collected all the URLs in our memory. Now phase two has already kicked in. We are now running the scraper. And here is the cool part. We are running 100 instances simultaneously. That loop node is set to batch size of 100. Meaning we are scraping 100 books details at the exact same time. You can increase or decrease this number based on your needs. But this makes the process incredibly fast. Processing hundreds of items still takes some minutes. So let's just skip to the finish line and we are done. Let's open up the CSV. Look at that. Hundreds of data points. We have the book names, the prices, the currency, availability, everything formatted perfectly. And remember, I still haven't written a single line of code. Now, before we wrap up, I want to show you the other two specialized modes this workflow has. First, SER extraction. If you need SEO data like tracking search rankings, just change the category in the form to search engine results. Second, manual mode. Sometimes you just want to inspect the code yourself. So just select general website and choose browser HTML. This gives you the row render dome. It's perfect for custom parsing or debugging. Or select screenshot to see exactly what the scraper sees. It captures the full visual render. This is a lifesaver for debugging layout issues or just verifying that the page loaded correctly. There are even more options in there to try out like network capture infight scroll but I will leave the rest for you to explore. And just like that we have upgraded from manually hunting for selectors to running an intelligent scraper that does the heavy lifting for us. Template and your three credits are waiting below. You build something cool with this or if you manage to break it, let us know in the comments. We had love to see what you're automating. Thanks for watching and I will see you in the next one. Happy skipping.