EN
Back to the archive

The archive · Commerce & Marketplaces · Product decision · 2024–2026

Malachyte's real-time intent bet: ex-Spotify engineers raise $10M for commerce AI

The team behind Spotify's recommendations bets online stores can read shopper intent live, without cookies, logins or purchase history.

Malachyte

The betReal-time behavior reveals intent before the first click, so cookie-free, history-free personalization will beat segment-based engines built on logins and past purchases.Scaling

What the business is

A behavior-intelligence platform that rewrites storefronts in real time: every hover, click, search and scroll sharpens a prediction of what the shopper wants next, with no account, cookie or purchase history required.

Starting capital$10M seed round co-led by Bessemer Venture Partners and Gradient Ventures, with Harpoon Ventures participating, announced 2026-08-06.

How it started

Sidd Motwani, Ian Anderson and Shivaditya Sinha spent years building the behavioral-intelligence system behind Spotify's recommendations - infrastructure TechCrunch says powers about 90% of recommendations to 800 million users. Watching retailers collect hovers, clicks, searches and scrolls without acting on them in the moment, they founded Malachyte in 2024 to bring the same intent-aware system to commerce.

What happened

Malachyte tested with more than 20 enterprise customers across travel, grocery and retail before focusing on e-commerce, went live with Fun.com in fall 2025, and reached general availability for Shopify merchants through a native integration in June 2026, with an API for larger retailers. Announced results include +31% revenue per visitor at HalloweenCostumes.com (Fun.com), an 80% lift in add-to-cart click-through at Brunt Workwear and +17% new-visitor revenue per visit at Jordan Craig.

How it ended up

Closed the $10M seed in August 2026 and is spending it on distribution, senior commercial hires and product leadership, with the platform live across a growing roster of brands.

Background

Malachyte is a New York-based behavior-intelligence startup founded by Sidd Motwani, Ian Anderson and Shivaditya Sinha, the engineers behind Spotify's recommendation infrastructure. TechCrunch reports their system, called Vector AI, powers about 90% of Spotify's recommendations to its 800 million users by predicting intent and next actions rather than relying on past behavior alone.

The founding bet is that e-commerce personalization is broken at its core: most stores treat first-time visitors as anonymous and existing shoppers as their purchase history, even though every hover, click, scroll, search refinement and add-to-cart is a live signal of what someone wants today. Malachyte's 'two-headed Vector AI' reads preference and intent simultaneously, forming a profile before the first click and sharpening it through the session, with no logins or cookies.

The company developed and tested the technology with more than 20 enterprise customers across travel, grocery and retail from 2024, launched with Fun.com in fall 2025, and reached general availability for Shopify merchants in June 2026 through a native integration, with an API for larger retailers. Customer-reported results at announcement include +31% revenue per visitor at HalloweenCostumes.com, an 80% lift in add-to-cart click-through at Brunt Workwear, and +17% new-visitor revenue per visit at Jordan Craig, where 80% of traffic is first-time.

In August 2026 Malachyte closed a $10M seed co-led by Bessemer Venture Partners and Gradient Ventures with Harpoon Ventures participating. The founders argue that as AI agents begin to shape buying journeys, commerce infrastructure that understands intent in real time - for humans and agents alike - becomes the bottleneck, and they plan to scale distribution while hiring commercial and product leadership.

What has to be true

  • The cold start is the biggest unserved hole in commerce: most personalization stacks can't recognize a first-time visitor at all, while acquisition costs keep rising.
  • The founders proved the core model at Spotify, where intent prediction from behavior - not just history - now drives the majority of recommendations at hundreds of millions of users.
  • Anonymous in-session signals give the system an unfair data advantage: every action sharpens the vector, so the experience improves the longer someone stays without identity data.
  • The bet extends beyond stores: if agents will increasingly shop on behalf of people, engines that read intent in real time become infrastructure for both humans and agents.

What can be applied

A hard-won model transfers better than code: Spotify's behavior engine became a commerce product because shopping shares the same cold-start math of predicting intent without history.

Aftermath

As of 2026-09-06 Malachyte is live with brands including Fun.com, Brunt Workwear and Jordan Craig, generally available to Shopify merchants since June 2026, and scaling after an August 2026 $10M seed from Bessemer and Gradient. Headline numbers are customer-reported at launch, so the open question is whether real-time intent personalization produces durable lift and converts the larger retailers the API targets. The Spotify pedigree and cold-start wedge give it a distinctive angle in a personalization market dominated by history-based engines.

Sources

spotted an error? The archive wants to know.

Your turn

You just read one. Describe what you are building, and see who is betting on the same thing.

Free account · 3 free questions · no card

Related cases