Here’s a demo I built using generated company descriptions and bull and bear cases for roughly 860 names disclosed by 25 funds I admire (Altimeter, Lone Pine, Coatue, etc.), embedded with Voyage’s voyage-4 model. To query that index, I built an agent running on Opus 5, with the fund universe serialized and held in a cached standing context. As a visual, I projected those same Voyage vectors down to three dimensions and plotted every company to see what that investing landscape looks like (check it out!). Building projects like this makes me really excited about the opportunity to expand access to public market investing and help retail investors build more conviction. Think about how scattered market-moving signals can be. Imagine if everyday investors could pair their own investment intuition with an understanding of where some of the smartest investors in the world are actually putting their capital. Take that a step further: imagine if the entire public market could be similarly indexed and queried in natural language by any investor, regardless of technical acumen. I think models like this could be a meaningful step toward a more efficient market. Companies like @dubinvest have approached this problem of knowledge accessibility in a really interesting way, opening up access to information and ideas that historically haven’t been easy for everyday investors to find. Would be excited to talk to anyone building in the space!