Ex-Spotify AI Engineers Launch Malachyte to Personalize E-Commerce in Real Time
Three former Spotify engineers raised $10M to bring intent-prediction technology from music streaming to online retail, enabling personalized shopping experiences without user accounts.
Last verified:
The Pitch: Intent Over History
Three former Spotify infrastructure engineers—Sidd Motwani, Ian Anderson, and Shivaditya Sinha—are applying behavioral prediction technology to online shopping. According to TechCrunch, their startup Malachyte closed a $10M seed round to scale the platform across e-commerce merchants. The founders spent years developing the machine-learning infrastructure powering approximately 90% of Spotify’s personalized recommendations across its 800 million users, and are now targeting a different supply problem: most online retailers show the same generic storefront to first-time visitors and rely on purchase history rather than immediate intent to recommend products to returning customers.
How Malachyte Reads Intent Without Accounts
Malachyte’s core innovation is the ability to predict shopper behavior from in-session signals alone. The platform ingests behavioral data—searches, clicks, dwell time, scroll depth, add-to-cart events—and constructs a shopper’s preference profile in real time, even before the first interaction. According to Motwani, a search for “heavy-duty boot” followed by two clicks on steel-toed footwear is sufficient for the system to elevate work apparel in the product feed and demote unrelated categories like dress shoes, with zero prior account or transaction context. The system uses what Malachyte calls “two-headed Vector AI”—one inference head for immediate purchase intent, another for taste profile—and updates both continuously as the user navigates.
Malachyte also weights contextual variables that most personalization engines ignore. A mobile visitor clicking an email link at 11 p.m. is in a different shopping mindset than the same user on a desktop browser at 9 a.m., but conventional segmentation treats both identically. By layering time-of-day, device type, and referrer source into the intent prediction, Malachyte narrows the customer experience to narrower behavioral states.
From Beta to Shopify Integration
The company piloted the technology across 20+ enterprise customers in travel, grocery, and retail starting in 2024 before narrowing focus to e-commerce. According to TechCrunch, the platform went live with Fun.com in fall 2025, and reached general availability via a native Shopify integration in June 2026. Larger retailers can integrate via direct API partnerships.
Why This Matters
Retail merchants have long treated personalization as a post-purchase or post-login challenge. Malachyte’s approach—inferring intent from session behavior without authentication—addresses a high-friction gap: the first-time visitor journey. If the intent-prediction model holds accuracy across diverse product categories and traffic patterns, this could reshape how mid-market and enterprise retailers structure their discovery funnels, potentially reducing reliance on paid acquisition to drive repeat visits. The Shopify integration lowers friction to adoption, though scale and retention across small-to-medium businesses will determine whether the $10M runway reaches unit economics.
Frequently Asked Questions
How does Malachyte differ from traditional e-commerce personalization?
Malachyte infers shopper intent from real-time behavioral signals (clicks, searches, hover time) without requiring login or historical purchase data. Traditional systems rely on segmentation, demographics, or past transactions, treating first-time visitors identically to returning customers.
What is 'two-headed Vector AI'?
According to Malachyte, the system predicts both immediate intent (what the user wants to buy now) and long-term taste (general preferences), refining both with each user action in the session and across repeat visits.
Where is Malachyte live today?
The platform launched with Fun.com in fall 2025, became available to Shopify merchants as a native integration in June 2026, and is also available to larger retailers via direct integrations.