Zagreb, Croatia · 2026IPCEI AI · Associated Partner
Sovereign AI Infrastructure
AI that
belongs
to the institution running it.
Congruence builds open R&D infrastructure for the EU AI ecosystem and production-grade sovereign systems for regulated industries, where AI must be certified, auditable, and independently controlled.
Thesis
Most organizations will never own their AI. That is a structural problem.
Congruence exists to close that gap. Not with another cloud-inference wrapper, but with a vertically-integrated stack that runs inside the institution's own perimeter. Compute, retrieval, reasoning, audit. One architecture. Two lines of output.
The Architecture
Two distinct lines. One foundation.
R&D LineOpen · IPCEI AI
Open research primitives for the European AI ecosystem.
Foundational R&D contributions developed under IPCEI AI, solving the hard infrastructure problems of sovereign, certified, and federated AI deployment. Released as open-source building blocks.
R04Certified UncertaintyProvable guaranteesEU AI Act
R05Neurosymbolic ReasoningRegulation as codeCertifiable
Commercial LineLive · Production
Production sovereign AI systems for regulated industries.
Purpose-built AI systems regulated institutions deploy on infrastructure they control. Framework-agnostic by design, extending to every sector where AI must be certified and owned.
Layered architecture. Production-grade modules. On-premise, air-gapped. Full audit trail on every AI decision. Human-in-the-loop at every lifecycle transition.
Any regulatory framework, natively, on infrastructure the institution controls, with zero cloud dependencies.
Congruence participates as an associated partner, contributing R&D work packages with demonstrable spillover across Member States and regulated sectors.
2026
II
EU AI Act First Mover
Fully enforceable in 2026.
Every high-risk AI system in Europe needs conformity assessment. Congruence is building the missing compliance layer, released as open source.
Article 15
III
Sovereign AI Factory Croatia
Sovereign GPU infrastructure.
Federated learning aggregation on auditable sovereign hardware. Physical infrastructure that makes European AI sovereignty operational, not theoretical.
SAF-HR
Team
Research-led. Production-grounded.
01
LP
Co-Founder
Luka Petrović, Ph.D.
Co-Founder
PhD in robotics. Computer Science Professor. Background in machine learning, probabilistic inference and production systems engineering.
02
VŠ
Co-Founder
Vlaho-Josip Štironja
Co-Founder
AI and robotics engineer. Background in LLM workflow engineering, real-time inference and estimation theory.
Memorandum
Request a technical briefing.
We demonstrate Congruence Sovereign against your specific regulatory framework and infrastructure requirements.