Moving Beyond Historical Data

Most e-commerce personalization relies on static data: past purchases, demographic segments, or logged-in profiles. Malachyte, founded by former Spotify engineers, argues this approach fails to address a shopper's immediate needs. Their platform, Malachyte, replaces these static models with "Vector AI," a system designed to predict intent and next actions in real-time without requiring a user account or historical data.

Real-Time Intent Prediction

Malachyte’s "two-headed" AI architecture processes behavioral signals as they occur. By analyzing micro-interactions—such as hovers, scroll depth, search refinements, and add-to-cart actions—the system continuously updates a user's intent vector.

For example, if a visitor searches for "heavy-duty boots" and clicks on steel-toed options, the system immediately re-ranks the storefront to prioritize work pants and gloves while deprioritizing dress shoes. This happens within a single session, allowing the experience to become more relevant the longer a user stays. The system also incorporates contextual signals, such as time of day, device type, and referral source, to differentiate the intent of a mobile user browsing at 11 p.m. from a desktop user browsing during business hours.

Implementation and Scaling

After testing with over 20 enterprise customers across travel, grocery, and retail, Malachyte launched general availability for Shopify merchants in June 2026. Larger retailers can integrate the technology via API. The company recently secured $10 million in seed funding, led by Bessemer Venture Partners and Gradient, to scale distribution and expand its product and commercial teams. The long-term goal is to unify merchandising and marketing efforts under a single, shared understanding of real-time customer behavior.