Enterprise AI Sovereignty with Apertus 1.5
Swiss businesses can now operate Enterprise AI on an end-to-end sovereign Swiss AI Technology Stack: Swiss Cloud + Swiss Model + Swiss Applications & Transactions
We are excited to announce that we are expanding our Enterprise AI platform by adding Apertus 1.5, the Swiss large language model developed by ETH Zurich, EPFL, and the Swiss National Supercomputing Center (CSCS). The combination of a Swiss-based infrastructure, a Swiss-based AI model, and a Swiss-based Enterprise Application Suite completes the sovereign AI technology stack for Swiss and European customers.
Apertus 1.5 is a fully open large language model developed within the Swiss AI Initiative and trained on the Alps supercomputer at CSCS in Lugano, Switzerland. The model is released under the Apache 2.0 license. It provides transparency into its development, with an open-weight, open-architecture, and open training process.
High LLM Selection Granularity with Audit Trail
For BlueCallom and its customers, the significance goes beyond adding another LLM. It makes AI sovereignty an architectural choice, at the AI-unit level. The latest BlueCallom platform, AgenticBlue 2.5, introduces highly granular model selection. Each AI unit can access a different LLM or AI service. An “AI-Unit” is the smallest intelligence component, equivalent to a prompt, in an Enterprise Workflow Intelligence application. A business-critical workflow may operate entirely on Apertus and Swiss infrastructure. Another task may use a specialized international model when it requires a particular capability. Image generation, for example, could selectively access an external service. And sensitive operational processes and enterprise data remain within the sovereign environment. The existing EAI Audit Trail in AgenticBlue records every move the AI performs. These moves include model selection, if the application allows it to select autonomously.
The architecture therefore lets organizations decide where sovereignty is required and where access to specialized capabilities adds more value. Until now, model selection in Enterprise AI has primarily been driven by two criteria: performance and cost. We added a third criterion: strategic relevance. You can now select an AI model based on the purpose of an individual task. That can include its data sensitivity, required capabilities, cost, and sovereignty requirements. This changes the LLM’s role inside the enterprise.
End-to-End Sovereignty
With Apertus 1.5, Swiss-hosted infrastructure and BlueCallom·AI, enterprises can establish an AI architecture in which the entire AI Technology Stack operates within Switzerland. This includes: Swiss Infrastructure → Swiss Foundation Model → Swiss Enterprise AI Platform → Enterprise Workflow Intelligence. Yet, with that level of granularity, Enterprise AI no longer needs a single model strategy. It can be as flexible as needed. This makes Enterprise AI even more adaptable, while keeping every action fully controllable and transparent.
If you are up for a detailed exposure to what we exactly did and do:
Let us know, and we can share all the details
Get in touch to explore how our Enterprise AI Development Framework could help you.
Per email: http://bluecallom.com/inquiry or phone: +41 (44) 500-6480
