Prosus introduces new AI-based framework to enhance ecommerce operations

The company launches its Large Commerce Model (LCM), an AI system trained on extensive commercial data to optimise digital transactions and consumer interactions.

Fabricio Bloisi, CEO at Prosus.

Prosus, a global technology investor that builds and partners with consumer internet companies, announced the launch of a new AI-based framework for ecommerce operations, named Large Commerce Model (LCM). The model is designed to integrate multiple ecommerce functions, such as search, recommendation and campaign optimisation, into a single agentic system.

The LCM was developed using Prosus’s proprietary database of global commercial transactions, which includes interactions from over 500 million users and more than 10 trillion data tokens. The model is intended to support a wide range of ecommerce applications, from search improvement to personalised recommendations, while operating at a lower cost compared to conventional AI systems.

According to Prosus, the new system reduces operational expenses related to marketing and data processing. It also enables the delivery of personalised user experiences across different digital interfaces, including voice assistants, augmented reality devices, and connected vehicles.

New model architecture for ecommerce operations

Traditional ecommerce platforms rely heavily on keyword searches and product recommendation engines. The Large Commerce Model introduces a different approach by incorporating long-term learning and contextual understanding of consumer behaviour. It enables the platform to recognise user intent and refine its performance over time based on past outcomes.

“The LCM represents a new operational layer for ecommerce, integrating AI and data in a single environment,” said Fabricio Bloisi, CEO of Prosus. “This model will be made available to companies across our global portfolio, enhancing their existing technology frameworks.”

Euro Beinat, Global Head of AI at Prosus, added that the system processes over 200 billion data tokens daily, improving its understanding of consumer behaviour and supporting decision-making across multiple markets.

Development and implementation

The model was initially developed with iFood, a leading food delivery platform in Latin America and part of the Prosus Group, and later expanded in collaboration with other Prosus portfolio companies, such as OLX, eMag, and Despegar.

The LCM combines five main capabilities:

Domain-specific learning: Trained on billions of real transactions, providing contextual understanding of consumer behaviour.

Unified operation: Integrates multiple functions—search, recommendation, and marketing optimisation—under a single AI environment.

Intent-based analysis: Identifies user motivations and patterns behind purchasing decisions.

Shared insights: Enables cross-platform learning within the Prosus network of companies.

Automated execution: Allows direct operational actions, such as adjusting marketing strategies in real time.

Measured results

The Large Commerce Model is currently in production across several platforms. Reported outcomes include:

Push notifications: Fourfold increase in order rates compared to previous benchmarks.

Search and recommendation: 66% improvement in conversion for “you may also like” suggestions; 4.6% increase in pharmaceutical order conversions.

Scalability: Currently serving 20 million users and processing more than 200 billion data tokens per day.

Operational efficiency: Estimated to operate at approximately 1/60th the cost of comparable proprietary models.

According to Zülküf Genç, Director of AI at Prosus, early user feedback from iFood’s restaurant partners indicates greater engagement and responsiveness to notifications generated through LCM-driven systems.