I
The brain
AI-native platform.
Federated learning, clinical models, sovereign data. The common foundation for the three other pillars.
Architecture
Three layers, from governance to action.
Action
AI agents, drones and robots execute at the point of care, for clinician and patient.
Collaborative intelligence
Federated learning, clinical knowledge sync, anti-bias safeguards.
Governance & trust
Zero-knowledge proofs (ZKP), on-ledger consent, automated GDPR / AI Act compliance.
Models learn without ever moving patient data.
Agentic workspace
A sovereign agentic workspace.
MedicalCity unifies patient data (FHIR), runs clinical analytics and builds governed AI agents — with HDS, AI Act and EHDS compliance by design. For health and autonomy.
BAYON agentic MDM
The platform’s source of truth: unified master data and rules, probabilistic arbitration of ambiguous cases, every decision traced and auditable.
Skill registry + agent registry
Every agent is declared — identity, version, authorised actions — and every production-validated skill is capitalised. Audit-grade governance.
Document compliance
Continuous verification and remediation of records and regulatory documents, under human supervision.
Patient-risk surveillance
Weak signals aggregated continuously over FHIR data; the clinician is alerted — never an autonomous decision.
Case scoping
Instruction and pre-qualification of files (APA, MDPH, admissions): completeness, documents, routing.
Multi-tenant EHR integration
Heterogeneous electronic health records connected on a shared FHIR foundation, isolated per facility.
Federated learning
Patient data no longer travels.
Rather than centralising records, MedicalCity sends the model to the data. Each institution trains locally; only the learnings — never the raw data — are synchronised.
The model is distributed to each facility.
Training happens locally, on data that never leaves.
Only aggregated, anonymised gradients return to enrich the global model.
Data sovereignty
Built for trust, by default.
Hosted in France
Health data hosted in France, on HDS-ready infrastructure.
End-to-end encryption
Encryption at rest and in transit; keys held under sovereignty.
GDPR by design
Minimisation, traceable consent and right to erasure built into the platform.
AI Act compliance
Risk classification, audit logs and documented medical evals.
Technical roadmap
From whitepaper to production.
-
Whitepaper v2.0
Current foundation
3-layer architecture, federated learning and on-ledger governance specified.
-
2027 — v3.0
Industrialisation
Federated learning in production across institutions, clinically validated interoperable standards.
Downloadable PDF behind an email gate.