01 Cavela
The founder grew up across Asian trade centers — Malaysia, Hong Kong, Singapore, China — and has three generations of family sourcing behind him. That's not a background story. That's why 200,000 vetted factory relationships exist in this platform and not Alibaba's. Every e-commerce brand in America is scrambling to exit China right now, and Cavela is the only tool that makes the pivot possible in days, not months. Customers are landing below pre-tariff unit costs despite 145% China tariffs. That sentence should stop you.
Round Seed · $8.6M Total | Lead XYZ Ventures + Susa Ventures | Team Ex-Stripe, Google, Amazon, Primark Sourcing
02 Juxta
GPS has never worked indoors, and every attempt to fix it — BLE beacons, UWB, WiFi triangulation — costs $250,000 per site to install. Juxta deploys the same capability with zero hardware: upload a floorplan, wait a few hours, start tracking. The mechanism is inertial sensor fusion plus ML trained on synthetic movement data, running entirely on-device. Signed a deal with RC Mowers for offline robot navigation. Defense pilots in discussion. The claim sounds impossible until you realize every phone and robot already has the sensors — no environment needed, just software.
Round Seed · $3.2M Lead Lux Capita Team Northwestern
03 Astor
The co-founders built Nubank (scaled to 100M users) and Robinhood/Stripe (scaled to hundreds of millions of dollars in GMV) — and then built a product where retail investors can call their AI advisor at 2am and get fiduciary guidance on their actual portfolio. Not a chatbot with disclaimers. An SEC-registered advisor that reads your Robinhood positions and tells you whether to exercise those RSUs now or wait. 3,000 users connected $60M in assets pre-launch. The robo-advisor graveyard is real — but none of them had LLMs, and none had this team.
Round Seed · $1.9 M | Lead Y Combinator | Team Ex-Nubank, Stripe, Robinhood
SIGNAL CHECK
The three picks this week share a structure: each one looks like a category that already failed. Procurement software. Indoor positioning. Robo-advisors. Conventional pattern-matching filters all three out. The actual signal is that each prior attempt failed because of a missing ingredient — generational trade relationships, on-device ML on commodity sensors, LLMs capable of portfolio-specific advice — and each ingredient only became available in the last 18 months. The bet isn't that these companies will succeed despite being in dead categories. It's that the categories only looked dead because the enabling layer wasn't ready.