Prosus Launches Privacy-First AI Agent Platform for European Businesses

The artificial intelligence revolution has largely been a story of massive foundational models built by American and Chinese tech giants, leaving European businesses grappling with a difficult choice: adopt powerful tools and risk exposing proprietary data, or prioritize privacy and fall behind. On Tuesday, European technology investment group Prosus introduced a third option with the launch of ToqanClaw, a privacy-first AI agent platform designed specifically to navigate the continent's strict regulatory environment.[1][2]

ToqanClaw functions as a "no-code" builder that allows anyone—regardless of technical background—to create custom applications, dashboards, and automated workflows simply by describing what they need in natural language. Built in-house and integrated with Prosus's existing AI infrastructure, the platform brings the capabilities of advanced autonomous agents into a secure, ring-fenced environment.[2][4]

The critical differentiator for European markets is the platform's data architecture. Prosus guarantees that all business data processed through ToqanClaw remains entirely under the user's control and is never utilized to train third-party foundational models. This addresses a primary hesitation among European enterprises, which operate under the stringent requirements of the General Data Protection Regulation (GDPR) and the newly implemented AI Act.[1][2]

"Everyone has a good AI model now. That's not the advantage anymore," said Prosus CEO Fabricio Bloisi during the launch announcement. "The question is who has the data, the context, and the loops that make it actually useful for a real business." Bloisi noted that the company spent 18 months developing the system internally before opening it to external partners.[2][3]

Rather than relying on a single underlying algorithm, ToqanClaw utilizes a dynamic routing system that evaluates each user request and directs it across more than 20 different AI models. This intelligent routing ensures that simple tasks are handled by faster, cheaper models, while complex analytical requests are sent to more robust systems, significantly reducing operational costs for end-users.[2][4]

The platform is currently rolling out in beta to Prosus's massive ecosystem of more than 5 million partners, a network that primarily consists of restaurants, independent merchants, and small entrepreneurs. For many of these businesses, hiring a dedicated engineering team to build custom software is financially impossible, effectively locking them out of the AI productivity boom.[2][3][4]

Early pilot programs have demonstrated substantial operational shifts for small and medium-sized enterprises. Lebkov & Sons, a Dutch café chain, utilized ToqanClaw to build an agent that reduced their financial reporting process from several weeks to just 30 minutes, a streamlining that management credited with supporting a 40% year-over-year revenue growth.[2][4]

Similarly, Burger & Frites, a Rotterdam-based restaurant, used natural language to construct a delivery analytics agent. The custom tool optimized their dispatch routing, resulting in a 25% increase in total deliveries, a 60% reduction in staff overtime, and an estimated €21,000 in monthly operational savings.[2][4]

Another pilot participant, the Dutch poke bowl chain Poke Perfect, created an operations assistant accessible directly via WhatsApp. The agent handles routine staff queries regarding scheduling, inventory, and procedures, reducing manager interruptions by 70% and allowing teams to focus on front-of-house customer service.[4]

The public launch of ToqanClaw follows an extensive period of internal testing at Prosus. The company has already deployed more than 60,000 active AI agents across its own global portfolio, automating complex workflows for its 15,000 employees. At iFood, Prosus's Brazilian food delivery subsidiary, a single senior agent now automates restaurant performance reporting, performing the manual equivalent of 40 full-time employees.[2]

To power these specialized business applications, Prosus trained its own Large Commerce Model (LCM) using anonymized data from more than 1 billion customers and 500 million daily interactions across its global network. When connected to ToqanClaw, the LCM allows the AI agents to move beyond simply executing requested tasks and begin proactively anticipating business needs before a human manager identifies them.[4]

The deployment represents a significant milestone in Europe's broader strategy to achieve technological sovereignty. While the continent may lag in developing the underlying foundational models, platforms like ToqanClaw suggest that Europe's competitive advantage may lie in the secure, compliant application of AI technology at the distribution layer, embedding intelligent agents directly into the daily operations of millions of small businesses.[1]