Albatross Real-Time Discovery

Albatross is a Zurich-based AI company founded in 2024 by former Amazon AI leaders, according to Retail Tech Innovation Hub. It builds real-time product discovery models for marketplaces and retailers that must show the right items as shopper intent changes within a session.

The platform uses transformer models over sequences of user interactions and catalog embeddings. It is served through one API with Discovery Feed, Catalog, Event and Prediction endpoints. The site reports use on marketplaces in 12 countries, with Wallapop and OfferUp named as customers, and offers a 7-week evaluation.

• Discovery Feed: returns a ranked list of items to show next for each user.
• Catalog API: ingests catalog embeddings for large product sets.
• Event API: captures real-time user interaction events.
• Prediction API: scores items using live user embeddings.
• Multimodal Search: adapts search results to shifting intent.
• Research-backed models: published work on cold-start recommendation.

• Vietnamese marketplaces can personalise homepage feeds in real time.
• Second-hand and classifieds platforms can surface new listings to likely buyers.
• Large retailers can adapt search results to in-session intent.

Marketplaces can personalise homepage feeds from in-session behaviour without building recommendation models in-house. The site reports Wallapop homepage engagement up 119 percent and OfferUp up 84 percent, company-reported results. A single API reduces integration effort.

Shoppers expect feeds that adapt instantly, and marketplaces compete on engagement. Vietnamese marketplaces and classifieds with large, fast-changing catalogs face the same cold-start and relevance problem.

Zurich, Switzerland, founded 2024. USD 12.5 million seed round in November 2025 led by MMC Ventures; total funding USD 16 million.