Working on Strot - an AI agent that reverse-engineers website APIs for scraping.
Instead of DOM scraping, it intercepts AJAX calls and figures out which API endpoint gives you the data you need. Uses visual analysis + fuzzy matching to identify the right call.
The use case: scraping product reviews, paginated listing data (products), etc. Existing AI scrapers either didn't work or were very slow and costly. A product with 1000 reviews takes 10+ minutes with Playwright, costs $10 with LLM scrapers. With Strot? 10 seconds via direct API calls.
Being used in production by a couple of clients. Would love feedback!
Very neat - I imagine you could even use this as a web scanner to identify security misconfigurations in API implementations (e.g. broken access control)
I actually enjoy refactoring messy code ->and have worked and improved multiple code bases in my time. However, I don't like recording videos. Any other alternative to apply.
I built a large remotion powered video generation pipeline for a major AI video company and one of the major challenges was cost of generating a video. 1 instance of Remotion at best generates 1-2 fullhd frame per second on server and that turns out to be quite expensive for slightly larger videos. From my benchmarking, taking screenshots was the slowest part of their pipeline. Any sense on how many frames revideo can generate. Since you could do canvas.toBlob, I am wondering if it's faster.
It really depends on the complexity of the scene. We're only getting started on optimizing rendering and I'm convinced that we can get it to be significantly more performant/cheaper than Remotion. Right now we're faster in some cases and slightly slower in others. Feel free to contact me at konsti at re.video. I would love to understand your use-case and see if and how we could help you.
About Us: Small software studio helping funded startups in solving their complex business needs - whether that involves transforming legacy systems, mastering video rendering and streaming, or scaling sophisticated AI applications
Lovely article and loved the analogy of home cook to making an app. Being a professional programmer who loves programming, I never thought about programming like this. But there is a catch here, I would wager that trying cooking at home is far easier and accessible as compared to making simplest of apps. Most of the no code low-code tools are focused on helping companies make software for their use and not focused on individuals making apps for themselves.
About Us: Small software studio helping funded startups in solving their complex business needs - whether that involves transforming legacy systems, mastering video rendering and streaming, or scaling sophisticated AI applications
We are a premiere software studio based out of India. We specialize in delivering high-quality software systems to the complex engineering problems of our clients. We have a strong background in web and mobile application development and have worked with startups large & small, remote & on-site around the world. We are passionate about open source, small high quality teams and seeing the business impact of software.
Instead of DOM scraping, it intercepts AJAX calls and figures out which API endpoint gives you the data you need. Uses visual analysis + fuzzy matching to identify the right call.
The use case: scraping product reviews, paginated listing data (products), etc. Existing AI scrapers either didn't work or were very slow and costly. A product with 1000 reviews takes 10+ minutes with Playwright, costs $10 with LLM scrapers. With Strot? 10 seconds via direct API calls.
Being used in production by a couple of clients. Would love feedback!
Blog: https://blog.vertexcover.io/strot-is-a-api-scraper GitHub: https://github.com/vertexcover-io/strot