Hi HN! I’ve been experimenting with large language models for data analysis and thought today’s demo day would be a fun way to gather feedback.
The problem I see is that for the average investor, it’s tough to get familiar with 100+ startups in a short amount of time. To make it easier, I scored the startups across 12 dimensions so anyone can filter based on their own investment thesis. There’s also a ranked search and similarity view, which makes it simple to focus on a specific industry or discover related startups.
On the technical side, the implementation is straightforward. I use a small model to score the startups and then chart them in two dimensions with PCA. Each score has a rationale attached, making it easy to audit or test. From my experiments, Gemini did a surprisingly good job with the scoring, which makes the tool both useful and fun for exploring startups.
Congratulations to all the teams launching today—there are some amazing and inspiring projects. Looking forward to your feedback!
The problem I see is that for the average investor, it’s tough to get familiar with 100+ startups in a short amount of time. To make it easier, I scored the startups across 12 dimensions so anyone can filter based on their own investment thesis. There’s also a ranked search and similarity view, which makes it simple to focus on a specific industry or discover related startups.
On the technical side, the implementation is straightforward. I use a small model to score the startups and then chart them in two dimensions with PCA. Each score has a rationale attached, making it easy to audit or test. From my experiments, Gemini did a surprisingly good job with the scoring, which makes the tool both useful and fun for exploring startups.
Congratulations to all the teams launching today—there are some amazing and inspiring projects. Looking forward to your feedback!