In 2012, a decade before generative AI became mainstream, I built a rule-based content-ranking engine — Blurbity — designed to surface the signal in any article.
Blurbity successfully used a rule-based system: position score, complexity score, keyword signals, etc., that surfaced the most important sentence in all the recent articles written about a specific topic and showed them together in a quickly scannable list.
Blurbity was selected to pitch on stage at LAUNCH Festival 2013 in San Francisco — Jason Calacanis's Techcrunch Disrupt spinoff for early-stage startups.
The problem it solved — surfacing what matters inside a body of information — is the same problem that now drives clinical summarization tools, AI research assistants, and large language models.
Building this in 2012 forced a rigorous answer to a question that's now embedded everywhere: what does "important" actually mean for content?