Former Spotify Employees Raise $10 Million for AI E-Commerce Startup

Former engineers who created Spotify’s music recommendation system founded Malachyte and raised $10 million in seed funding to scale its Vector AI technology for online shopping. TechCrunch reports that the funds will be used to expand product distribution and hire new specialists. Techcrunch.com reports .

Sidd Motwani, Ian Anderson, and Shivaditya Sinha previously helped develop Spotify’s behavioral intelligence infrastructure. The system, called Vector AI, is designed to predict users’ intentions and next steps rather than relying solely on their past actions. It currently powers nearly 90% of recommendations on the music platform, which has more than 800 million users.

Problems with Traditional Online Stores

According to Malachyte’s founders, most online stores today serve shoppers with the same basic approach. Personalization largely relies on purchase history, demographic segmentation, or profiles of logged-in users.

As a result, first-time visitors see a generic storefront that is identical for everyone, while returning customers receive recommendations based mainly on what they bought before rather than on what they need today. The new startup aims to solve this problem by creating a shopping experience that understands each customer’s intent in real time.

The Capabilities of Two-Headed Vector AI

The company’s platform uses “two-headed Vector AI” to predict what product a shopper may want next, learn their broader preferences, and continuously improve based on real-time actions.

In an interview with TechCrunch, Malachyte CEO Sidd Motwani said the system starts working as soon as a page loads, using the available context. According to him, a single session is enough to determine a user’s preferences and what they are trying to achieve at that moment.

For example, a search for “durable boots” and two clicks on a steel-toe footwear page are enough to move workwear and gloves higher on the page. No account or browsing history is required, and every additional action makes the profile more accurate.

The company’s leadership also notes that retailers possess the most valuable data about their customers but rarely use it in real time. Every cursor movement, click, search refinement, and cart addition is continuously analyzed to identify user intent more reliably.