How I built a browser automation workflow by just describing what I needed—no coding

[AzureNova]

I’ve been stuck for months trying to figure out how to automate some repetitive scraping work. Every time the website changed its layout even slightly, my scripts would break and I’d have to rewrite everything from scratch. It was honestly draining.

Last week I tried something different. Instead of writing out selenium scripts line by line, I just described what I wanted to happen in plain English. Basically told the system: “log into this site, grab the product prices, extract the titles, and dump it into a spreadsheet.” Nothing fancy, just what I actually needed.

Turned out the AI could translate that into a working workflow that actually handled the changes without me touching the code again. When the site changed some CSS classes, the workflow adapted instead of just dying.

I’m curious though—has anyone else tried this approach? How reliable has it been for you when sites actually went through real UI overhauls, not just minor tweaks?

[bronze_boreal]

I’ve been doing this exact thing for about a year now and it’s genuinely changed how I approach automation. The biggest win for me was when a client’s vendor completely redesigned their dashboard. My old script would’ve taken days to fix.

Instead, I just re-described what I needed in plain text and the workflow regenerated itself. The AI understood the intent behind what I was doing, not just the specific selectors.

The key is being clear about what you’re trying to achieve, not how you’re trying to achieve it. Let me share what made this work so well for me—I stopped thinking like a developer and started thinking like someone describing a job to another person.

You can explore this more and see templates other people have built at https://latenode.com