Shaped

Shaped, a New York-based AI startup founded in 2021 with $10M in funding, has been acquired by Whatnot. The entire founding team (Tullie Murrell, Daniel Camilleri, Shai Bruhis) and core engineering staff are transitioning to Whatnot.

Leadership read: Whatnot was already running ML-based recommendation systems, but this acquisition commits it to something operationally distinct: a dedicated applied AI research function, led by Shaped's founding team, focused specifically on real-time ranking in a live-inventory environment. That is different from iterating on existing collaborative filters. The cross-category growth figures Whatnot cites (170% year-over-year) mean the discovery surface is expanding faster than a standard recommendation architecture can track; Shaped's sub-50ms inference infrastructure was built precisely for that kind of session-level, multi-sided relevance problem. Whatnot has now absorbed a production-grade technical capability and the team that built it. The related signal set here is broad M&A activity across twelve transactions in the last 90 days, spanning enterprise networking, real estate, defence, and fintech. Honest read: none of the comparable signals map closely to this deal's thesis. The relevant pattern sits narrower, in acqui-hire activity at the intersection of applied AI and marketplace infrastructure, where platforms are pulling ranking and personalization talent in-house rather than buying commercial API access. That is a distinct strategic posture and it is becoming more common among scaling commerce platforms. Across companies at this stage of marketplace growth, the pattern surfaces consistent functional pressure in applied ML and real-time systems engineering, product leadership capable of managing multi-sided relevance tradeoffs, and research operations that can bridge academic ML and production deployment at speed.

curated · 2026-07-15 · context →