EVERYTHING
Former Spotify engineers secure $10M for real-time commerce AI
The creators of the recommendation infrastructure behind 800 million active Spotify accounts have officially launched their new enterprise. Startup Malachyte announced a $10M seed funding round co-led by Bessemer Venture Partners and Google's Gradient Ventures, alongside Harpoon Ventures. Co-founders Sidd Motwani, Ian Anderson, and Shivaditya Sinha previously developed Vector AI, the proprietary engine responsible for generating roughly 90% of Spotify's music recommendations.
Rather than relying on past listening habits, Vector AI parsed real-time user behavior to determine current mood and context. Malachyte is now applying this identical architectural philosophy to global digital retail, aiming to eliminate the static recommendation widgets that have dominated online shopping for over two decades.
Replacing outdated profiles with continuous vector math
Traditional e-commerce platforms struggle with a fundamental limitation: they treat every visitor either as a blank slate or as a static profile defined by past orders. Malachyte replaces this approach with what it terms two-headed Vector AI, a continuous inference model that processes signals before a customer even completes their first click.
The system evaluates multiple micro-behaviors simultaneously during a single active browsing session:
- Page entry context, including time of day, referral links, and device type
- Real-time interaction telemetry such as hover duration, scroll speed, and image gallery swipes
- Session-specific search refinements and immediate cart additions
«A search for 'heavy-duty boot' followed by two clicks on steel-toed boots is enough to move work pants and gloves up the page and push dress shoes down, with no account or history required,» explained Sidd Motwani, chief executive officer of Malachyte, in an interview with TechCrunch. «Every additional action sharpens the profile, so the experience gets more relevant the longer someone stays.»
Why intent prediction redefines the future of online retail
The broader implications of instant behavioral modeling extend far beyond higher click-through rates on product pages. By shifting from historical transaction logs to real-time intent prediction, retailers can dynamically restructure entire storefront layouts on the fly, effectively giving every shopper a custom digital store tailored to their immediate task on 6 August 2026.
This approach solves the privacy dilemma facing modern e-commerce. As browser cookies fade and privacy regulations tighten globally, algorithms that read real-time session context eliminate the need to track users across the web or force account creation. Following initial deployments with Fun.com and native integration for Shopify merchants, the technology proves that AI can deliver hyper-personalized experiences while respecting consumer privacy—signaling a structural shift toward context-driven digital commerce.