European AI decision model Metis challenges US dominance

A different approach to AI decision-making

While many AI tools generate lengthy text responses, Metis takes a different approach. The model answers yes/no questions, multiple-choice queries, and scale-based assessments by weighing every possible answer and providing a confidence score for each. This structured output allows companies to integrate AI decisions directly into their existing systems without having to parse unstructured text.

For instance, when analyzing a customer support ticket about payment failures, Metis simultaneously determined whether a live service was down (95% confidence), which team should handle it (technical team, 93%), and the urgency level (immediate, 96%). The entire analysis took 72 milliseconds and consumed just 0.0047 watt-hours of energy.

"Businesses make thousands of decisions every day, and many involve choosing between fixed options rather than composing explanations," says Robert Keus, spokesperson at GreenPT. "Metis gives you probabilities you can act on immediately for clear cases, while flagging uncertain situations for human judgment."

Speed meets sustainability

The model's compact architecture requires significantly less computational power than text-generating systems. By reading all input once and calculating probabilities in a single step, rather than generating tokens sequentially, Metis delivers responses up to five times faster than comparable chat-based systems. Every API response includes an impact report detailing the energy consumed, allowing organizations to track their environmental footprint per request.

Data sovereignty and renewable energy

Unlike many AI decision models currently available, Metis operates exclusively from GreenPT's EU endpoint. All processing occurs within European data centers powered by renewable energy sources, addressing growing concerns about data sovereignty and environmental impact in the AI sector. The company does not use customer data for training purposes.

The market opportunity appears substantial. American competitor TypeSafe AI reportedly reached an annual revenue run rate of $100 million within seven days of launching their decision model Jev, indicating strong demand for this category of AI tooling.