Bridging the gap between generative AI and medical accuracy through hybrid local/cloud inference. We solve LLM-hallucinations via a deterministic RAG pipeline.
Körperfluss builds a neuro-symbolic AI-middleware for medical education in the DACH region. We solve LLM-hallucinations via a hybrid RAG-architecture combining local NPU inference for privacy with Google Vertex AI for scalable reasoning. Seamlessly integrated into university LMS via LTI 1.3, we provide safe, deterministic clinical training. Validated by FFG Funding #938346.
Our neuro-symbolic logic engine ensures that generative outputs remain grounded in academic medical standards.
To ensure seamless integration into existing university ecosystems (e.g., Moodle), we deploy our LTI 1.3 microservices on Cloud Run, guaranteeing zero-downtime scaling during peak exam periods.
While sensitive patient data is processed locally via M4 NPU inference (Edge AI) to ensure strict GDPR compliance, the heavy lifting of academic inference, embedding generation, and global RAG orchestration relies entirely on Google Cloud's robust and secure infrastructure.