The world's largest database of new website launches
This report uses machine learning to identify genuinely novel business concepts that most people haven't heard of yet. We're looking for the next "rage room" - experimental ideas that represent new human needs or cultural shifts.
We analyze embeddings (AI-generated semantic representations) of recently launched websites using HDBSCAN clustering to find groups of similar businesses.
Each cluster is compared to all historical business patterns using cosine distance. Clusters with >0.5 novelty score are genuinely different from anything we've seen before.
We prioritize tiny clusters (2-5 sites) because they represent early emerging trends before they become mainstream. Even single novel sites are worth tracking.
GPT-4 analyzes each cluster on 5 dimensions: Concept Novelty, Cultural Shift, Cross-Category Innovation, Might-Fail Factor, and Entrepreneurial Signal.
Only concepts scoring 5.0+/10 overall are included. We filter out established categories (like restaurants, salons, etc.) by cross-referencing against 698 known business types.
For Entrepreneurs: These signals represent potential market opportunities before they become crowded. A small cluster size (2-5 sites) indicates you're early to the trend.
For Investors: High "Might-Fail" scores indicate experimental concepts worth watching. Cultural Shift scores reveal changing consumer behaviors.
For Researchers: This data tracks real-time cultural and economic shifts through the lens of entrepreneurial activity.
Note: Not all signals will succeed - many might fail. That's part of the value: seeing what entrepreneurs are experimenting with in real-time.