Clinical data stays in the organisation
On-premise or encrypted hybrid cloud deployment, with LLMs run locally. Clinical data does not leave the organisation’s perimeter: data control is in the architecture, not in a promise.
AgenticHealth
A clinical data platform built for the organisation's perimeter: local data, AI by the doctor's side and every interaction logged are part of the product.
Clinical data with clear accountability
AI by the doctor’s side
AgenticHealth’s agents propose, document and cite sources. The diagnosis, the clinical decision and the signature stay with the doctor: human-in-the-loop and clinical accountability by design.
Every AI interaction logged
Full audit trail of every interaction: which document was retrieved, which source cited, what the model answered and who validated. Ready for clinical validation and the MDR pathway.
Compliance and data
The answers procurement asks for first
MDR planned, RUO today
Diagnosis support is a Software as a Medical Device function: development follows the MDR pathway (Reg. EU 2017/745). The launch is in Research Use Only (RUO) mode, pre-CE.
GDPR and professional secrecy
Special-category health data processed under GDPR (arts. 9 and 28) and professional secrecy. Patient data is not used to train shared models.
Interoperability on standards
EHR/HIS connectors and open standards: HL7 FHIR, DICOM, terminologies SNOMED CT, ICD-10 and LOINC. Data stays interoperable and traceable to its sources.
Frequently asked questions
Does clinical data leave the organisation?
No. AgenticHealth is built for on-premise or encrypted hybrid cloud deployment, with LLMs run locally (including medical models). Clinical data stays inside the organisation’s perimeter: no calls out to external services.
Does AgenticHealth diagnose in place of the doctor?
No. It proposes differential diagnoses with the rationale and cited sources, as support. The diagnosis, the clinical decision and the signature stay with the doctor: it is designed human-in-the-loop, with clinical accountability held by the professional.
Do you use patient data to train the models?
No. Patient data serves to answer on that patient’s case and to build their digital twin, not to train shared models. Processing under GDPR and professional secrecy.
Is it a CE-marked medical device?
Not yet. Diagnosis support is classifiable as Software as a Medical Device and development follows the MDR pathway (Reg. EU 2017/745). The launch is in Research Use Only (RUO) mode, pre-CE.
Where do the chatbot’s answers come from?
From your own data and documents: reports, lab results, discharge letters and imaging reports indexed via RAG (HL7 FHIR, DICOM, clinical PDFs), and from guidelines and literature for diagnosis support. Every answer links back to the sources it is drawn from.
Does it integrate with our EHR and hospital systems?
Yes. EHR/HIS connectors and open standards (HL7 FHIR, DICOM) with SNOMED CT, ICD-10 and LOINC terminologies. Integrations are mapped to your case during activation.
How do remote follow-up and the digital twin work?
Post-discharge follow-up uses AI-guided questionnaires and IoMT data, with alerts to the clinician on worsening parameters. The digital twin is a longitudinal patient model on which to simulate therapy scenarios and estimate outcomes, always as decision support.
How much does it cost and how do we start?
With licences for healthcare facilities and research centres, tailored to facility, volumes and add-ons (digital twin, follow-up). Quote on request. We start from a conversation about your clinical and organisational case.