A Tinder-style lead platform that renders live website previews in a PWA. Swipe right to trigger an outreach email; swipe left to discard. A custom recommendation algorithm analyses the HTML structure of sites you liked and surfaces new leads that match your preference profile. Every interaction trains your pipeline.
Traditional lead qualification means opening site after site, manually evaluating each one. This platform makes it into a swipe-based interface — you see a live preview of the site, make a split-second decision, and move on. Right swipe triggers an AI-personalised outreach email instantly.
The more you swipe, the smarter the algorithm gets. It builds a profile of what "a good lead" looks like for you specifically — and starts surfacing recommendations before you'd ever find them manually.
When you like a site, the system parses its HTML structure — tag patterns, layout signatures, font choices, link density — and creates a feature vector. Your "liked" vectors are averaged into a profile. New sites are scored against this profile using cosine similarity. Sites above 0.78 threshold get queued automatically.
Rendering external websites inside an iframe-based preview without CORS blocking, mixed content issues, or broken layouts — using a server-side proxy approach.
Choosing which HTML features to extract that actually correlate with "site quality" — and normalising them across wildly different site architectures.
With fewer than 5 liked sites, the similarity engine has too little signal. Solved with a curated seed set of "known outdated" sites to bootstrap the profile.
Each outreach email needs to reference something specific about the site. The HTML analysis output feeds directly into the email template generator.