Wallapop partners with Swiss AI firm Albatross to roll out real-time, in-session discovery across its consumer-to-consumer marketplace, aiming to increase buyer engagement and improve visibility for sellers. The companies said the deployment shifts Wallapop away from recommendation systems that rely primarily on historical behavior toward models that adapt to user intent as it changes during a browsing session.
According to the announcement, the technology is designed to interpret the sequence of actions a user takes in real time and update recommendations continuously within the same session. Wallapop said this can change what users see on the homepage and throughout browsing as their intent evolves, rather than continuing to recommend near-identical items based on the first click.
The companies described examples in home-furnishing scenarios to illustrate the intended effect. If a user begins by viewing a second-hand sofa and then navigates toward rugs or floor lamps, the system is intended to infer a shift from evaluating a single product to furnishing a room, and to surface complementary categories instead of more sofas. If the user then clicks a vintage coffee table, Wallapop said the experience can pivot again to show items matching the inferred style, such as lighting, décor, artwork, storage, or related furnishings.
Performance claims were based on an A/B test and subsequent production rollout, the companies said. In a four-week test on 10% of Wallapop’s traffic, measured across the full homepage experience, Wallapop reported a 118.9% increase in user engagement, a 104.8% increase in favorites and interactions, and a 46.9% increase in purchase intentions. The announcement also said the results stayed stable and continued to improve over a four-month production period after the initial test.
Wallapop and Albatross argued that the approach is particularly relevant for recommerce marketplaces, where inventories are made up of one-of-a-kind listings and metadata can be inconsistent. In such environments, they said, popularity-driven and similarity-based recommenders can struggle to reliably match buyers to relevant items, especially when supply is fragmented and each listing is unique.
The companies also positioned the rollout as a seller-facing capability, not only a buyer personalization feature. Wallapop said “algorithmic seller discovery” can help match live buyer intent with relevant, shippable listings so that high-quality supply is less likely to be buried by volume, addressing a long-standing challenge in second-hand commerce where visibility can be uneven across sellers.
Rob Cassedy, CEO of Wallapop, said the company is working toward a system that understands what users want “in real time,” with the goal of helping buyers find items faster while improving listing visibility for sellers. He added that the partnership supports Wallapop’s stated mission around “conscious consumption” and creating economic opportunities for users participating in the marketplace.
Albatross framed Wallapop as a high-pressure testbed for scaling real-time discovery because of the marketplace’s fast-changing catalog and high listing velocity. Dr. Kevin Kahn, CEO of Albatross AI, said the collaboration aims to deliver a discovery experience closer to in-store browsing, where shoppers can find “hidden gems” as preferences shift, and characterized the technology as a model that interprets context from user actions as they occur.
The announcement follows Albatross’ recent $12.25 million fundraising, which the company said is intended to support deployment of its real-time “perception layer.” The companies described the Wallapop rollout as an early large-scale commercial deployment of adaptive, in-session AI discovery in recommerce, positioning it as a step in Wallapop’s broader strategy to advance marketplace search, discovery, and personalization.
