mprofi AI Hub runtime
Own AI runtime for the mprofi stack
What we are examining
When several Beherzig products use AI models, we need a shared runtime — self-hosted, Swiss servers, clear tenant separation. We are examining whether an own runtime is economically viable against external providers.
Hypothesis
— From five products on, the own runtime starts to pay off — Swiss data sovereignty is a non-negotiable argument for the target audiences — Open-source models suffice for most use cases
Status
Concept. Architecture sketch and cost model in Q3 2026.