Europe has a complicated relationship with American AI tools. The data sovereignty concerns are real, the regulatory exposure is significant, and yet most businesses on the continent have been quietly using platforms built and trained on infrastructure that answers to no one in Brussels. Prosus thinks it has an answer.

The Amsterdam-headquartered consumer internet conglomerate officially launched ToqanClaw on June 23, 2026, positioning it as the first scaled European no-code AI platform built with GDPR compliance and data sovereignty as structural features, not afterthoughts.

What ToqanClaw actually does

The platform lets users create applications, automate workflows, and generate dashboards without writing a single line of code. The target user is not a developer. It is a restaurant owner, a merchant, or an entrepreneur who needs operational software but does not have the budget or the staff to commission custom builds.

Prosus says ToqanClaw is aimed at its ecosystem of more than 5 million partners, spanning food delivery, retail, and payments verticals. The platform is currently in beta.

Euro Beinat, head of data science and AI at Prosus, confirmed that user data stays under individual control and is explicitly not used to train third-party models.

Built on existing infrastructure Prosus already owns

ToqanClaw did not appear from nowhere. It is the external-facing layer of a system Prosus has been developing internally for some time.

The underlying engine is Toqan, an AI agent platform already running across Prosus’s internal operations. More than 25,000 associates across 22 companies in the Prosus portfolio currently use the internal Toqan system, which processes approximately 747,000 actions every month.

ToqanClaw takes that infrastructure and opens it to external users. Launching simultaneously alongside it is Zapia, a consumer-facing AI assistant designed to extend usability for everyday users and integrate into the broader Prosus AI stack.

The technical architecture includes intelligent request routing, meaning queries get distributed across multiple AI models to optimize performance and resource efficiency. Beinat described this as enterprise-grade infrastructure made accessible.