Spotify’s Recommendation Engine Is Coming to Online Stores
The same behavioral intelligence that decides what 800 million Spotify users hear next is now being deployed in e-commerce. Three former Spotify engineers who built the streaming giant’s Vector AI recommendation system have raised $10 million in seed funding for Malachyte, a startup that aims to fix what online retail has largely ignored: the first-time visitor.
Malachyte’s platform uses what it calls “two-headed Vector AI” to predict what a shopper wants in real time—not based on what they bought six months ago, but on what they’re doing right now. The round was co-led by Bessemer Venture Partners and Gradient, Google’s AI-focused fund, with participation from Harpoon Ventures. The company has been developing the technology since 2024 and went live with its first customer, Fun.com, in fall 2025.
What Makes Malachyte Different
Most e-commerce personalization today relies on historical purchase data, demographic segmentation, or logged-in customer profiles. The result: first-time visitors see generic storefronts, and returning customers get recommendations based on what they already bought rather than what they actually need.
Malachyte’s system starts working before a visitor’s first click. It uses contextual signals available the moment a page loads—device type, time of day, referral source—and combines them with real-time behavioral data: hovers, clicks, scrolls, search refinements, and add-to-cart actions. The platform treats each interaction as a signal that sharpens its understanding of both the shopper’s general taste and their immediate intent.
“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,” CEO Sidd Motwani told TechCrunch.
Context also matters. Motwani points out that a phone visitor at 11 p.m. arriving from an email link is in a fundamentally different mindset than the same person on a laptop mid-morning—yet most systems treat them identically.
The platform is now generally available to Shopify merchants through a native integration, while larger retailers can integrate via API.
Why This Matters Now
E-commerce faces a structural contradiction. Advances in machine learning-driven ad targeting on Google, Meta, and TikTok have made audience acquisition more precise than ever. But the website experience most shoppers encounter remains largely static. Retailers are paying more to acquire customers—customer acquisition costs rose roughly 40% between 2023 and 2025—only to convert them with personalization systems that fail to recognize anonymous visitors.
The average e-commerce brand now loses $29 per new customer after marketing and returns. Every unconverted visit compounds that loss.
Malachyte’s pitch is that retailers already possess their most valuable source of customer intelligence—behavioral signals—but rarely act on it in real time. Most systems either ignore these signals entirely or aggregate them into overnight segments. The startup reads them continuously, building a profile that grows more accurate with every action.
The Technology Behind the Pitch
Under the hood, Malachyte uses a two-headed vector AI architecture. One component models slower-moving preferences—what the shopper generally likes. The other reacts to what the shopper appears to want during the current session. Both fine-tune continuously based on real-time actions.
The system is persistent across channels. A signal picked up during search sharpens what the same shopper sees later on a product page, a category page, or even in an email or SMS the next day. It operates at sub-200ms latency under Cyber Monday-level load, auto-scaling without outages. Merchants retain control through a merchandiser layer that allows rule-setting and overrides. Onboarding takes as little as seven days.
Importantly, the system works without cookies, logins, or pre-existing customer data. That’s a meaningful distinction in an era of tightening privacy regulations and declining cookie effectiveness.
The Cold-Start Problem Malachyte Solves
The “cold-start problem” is one of e-commerce’s oldest headaches. When a first-time visitor has no history, traditional personalization systems have nothing to work with. They default to generic best-sellers—exactly when a first impression matters most.
Malachyte’s approach reframes this problem. Rather than needing to know who the visitor is, the system identifies what products they interact with and draws on those products’ existing behavioral relationships to personalize from the very first click. This is a fundamentally different approach from the login-or-nothing model that dominates retail personalization today.
The company worked with more than 20 enterprise customers across travel, grocery, and retail before focusing exclusively on e-commerce. That cross-sector testing suggests the underlying technology is adaptable, though the startup has clearly identified retail as its primary market.
What the Investment Signals
Bessemer and Gradient aren’t backing a generic personalization tool. They’re betting on a specific thesis: that e-commerce conversion rates have flatlined around 2-3% industry-wide, even as digital ad costs keep climbing. The next wave of efficiency won’t come from better ad targeting—it will come from converting the traffic retailers already pay for.
Bessemer partner Maha Malik framed the opportunity in terms of the coming “agentic era” of commerce. By 2030, McKinsey projects that AI agents could orchestrate up to $1 trillion of U.S. retail. Yet most commerce infrastructure isn’t built to understand buyer intent in real time, whether for a human shopper or an automated agent. Malachyte is positioning itself as the behavioral intelligence layer for that future.
What’s genuinely interesting here isn’t just the Spotify pedigree. It’s that Malachyte is tackling the most stubborn problem in e-commerce personalization—the anonymous visitor—with a solution that doesn’t require trade-offs. Most personalization tools force retailers to choose between privacy (no tracking) and relevance (historical data). Malachyte claims to deliver relevance without the data trail. If that holds up at scale, it changes the economics of customer acquisition fundamentally. The $10 million seed round suggests investors believe the technical claim is credible. The real test will be whether the platform can maintain its personalization quality as it scales beyond its initial Shopify integration to larger, more complex retail environments.
What Comes Next
Malachyte plans to use the funding to scale distribution and hire senior product and commercial leadership. The immediate priority is expanding beyond its current Shopify footprint to reach more retailers through its API.
Motwani says the bigger opportunity is bringing merchandising and marketing together around the same understanding of customer behavior. Today, these functions often operate with different data and different incentives. A unified behavioral intelligence layer could bridge that gap.
For retailers, the calculus is straightforward. Customer acquisition costs are rising. Conversion rates are flat. The infrastructure for understanding shoppers in real time exists—it just hasn’t been applied to commerce at scale. Malachyte is betting that the same technology that predicts what 800 million people want to hear next can also predict what they want to buy.
