I run two small products I launched this year and as a daily task I was keeping an eye on my competitors. Two things that I was particularly curious about were: the new features they are launching and the content they are posting. I was initially checking their blogs manually and handpicked the topics I was curious about, then finding new angles, proper keywords. Same for the features - skimming the changelogs, looking at the homepages and searching for mentions of any new products.
This is a boring task that takes time. But I had to do it daily because I wanted to keep my products up with the competition.
So I started to automate it. A few cron jobs, a few parsers. But still, there was the manual part like reviewing the changes, and separating the signal from the noise. There was a lot of noise like cookie banners, CSS changes. But it was working decently and I was okay with that.
Later I got curious about the idea of tracking the pricing pages. And here is where my automation started to break. Compared to blogs and changelogs, the pricing pages have unpredictable formats. It's practically impossible to write a parser that will work reliably and get all the useful information. Here I tried to send the pricing page content to LLM, hoping that it will get the useful information and ignore all the noise. It wasn't cheap and wasn't reliable. I guess because LLMs are not deterministic and they can return different data from the same input. Garbage in, garbage out.
Then I started looking for any products that would help me fix all these problems and the first one I stumbled upon was Visualping. Pretty neat tool, but after a few days of using it - I realized that I'm not making my work easier. I still have to review the changes manually and decide what to do about them. This is the moment I realized that I may not be the only one who has this problem and I started digging deeper into my automation scripts.
It took me about a month to build something that I can actually use. I'm still working daily trying to get rid of as much noise as possible because the more websites I run through it - the more edge cases I find. But it's getting there, so I'm pretty optimistic. That's why I'm sharing it with you.
Here it is: https://competiflow.com
You paste a homepage, it discovers pricing/docs/blog/changelog URLs, re-reads them on a schedule, extracts fields like plan names and prices, and diffs those instead of raw pixels. When something real changes you get a short summary plus the before/after text and the source URL. If it's a blog post - it generates relevant keywords and looks at the search volume, so you know if this post is worth looking at. There's a REST API and MCP server too. I use the MCP from Cursor when I'm working on other stuff.
It has a free trial so you can test it out without spending any money. Happy to answer any questions.