Description: This specification defines a complete multi-agent orchestration architecture for small and medium enterprises, pharmacies, and service providers operating within DE/EU regulatory frameworks. The system is designed to run on commodity hardware without cloud, DevOps, or enterprise IT dependencies. Core components: trigger layer (email/calendar/file/manual), rule-based router with optional Bayesian scoring, typed agent registry (12 agents with formal I/O schemas, constraints, and privacy flags), state-machine workflow orchestrator, and a dashboard-first UI for non-technical operators. All final decisions require human approval (human-in-the-loop by design). Key capability: systematic coupling of enterprises with research institutions via anonymized contact packets, NDA templates, and funding opportunity matching — enabling structured knowledge transfer without exposing confidential business data. Includes: system architecture, data models (YAML), 12 agent definitions, 3 agent constellation configurations, 3 end-to-end demo scenarios, dashboard UX specification, GDPR-compliant security/privacy design, and a phased implementation roadmap (MVP → v1 → v2). Optional modules for Bayesian routing and Monte Carlo risk simulation are specified with clean interfaces. All configurations are implementation-ready and technology-stack-agnostic at the specification level. Keywords: multi-agent systems, agent orchestration, SME digitalization, research-industry coupling, human-in-the-loop, GDPR compliance, low-tech AI, workflow automation, knowledge transfer, regulatory compliance
W. H. Brueckner (Sat,) studied this question.