For university hospitals
One platform for patient care and research.
Clinic and research work in the same audited environment, on your hospital’s own computing power.
- On your GPU cluster
- Your own models
- Python interface
- Permissions from your directory
- Connection via FHIR
- ISO 27001, C5 in progress
Instead of AI islands in every department.
One access point, one permission system and one audit log for the whole hospital.
- GPU server at the department
- Private chatbot accounts
- Start-up app on its own server
- Wards without AI
- One cluster for everyone
- One access point with your permissions
- Apps audited in the App Store
- Patient care with AI
In patient care: AI for patient data.
As in every hospital with basebox, only larger.
- Everyone sees only their own data
- Every answer with a source
- Ready-made apps instead of an empty chat window
In research: your models, your scripts, your apps.
One access point for every model.
The one we tested, your own, or one already running on your cluster.
Details
You decide which model answers: the one we tested, your research group’s own model, or one already running on your cluster. For everyone who works with it, nothing changes: the same apps, the same permissions, the same audit log.
Via an interface from Python and Jupyter.
Existing scripts run with a changed address.
Details
Every researcher gets a personal key. The interface is OpenAI-compatible; existing scripts run with a changed address. Every call runs with the permissions of the person making it and is recorded in the audit log.
Your hospital’s own apps run audited on your server.
From start-ups or from your own staff, each as its own container.
Details
Hospitals produce more and more of their own apps, from start-ups and from doctors who build something for their speciality. basebox adds such apps to the App Store. We audit them for security; each then runs as its own container on your server, with everything it needs, such as its own database. The app uses basebox’s models, knowledge and connections through the interfaces. Permissions, sharing and the audit log apply as for any other app. ibasebox runs on Kubernetes. Every service is its own container, separate from the others.
Use cases for research and the clinical trials office.
Research and patient care stay separate.
Researchers see only what your hospital releases for research.
- Own groups and own knowledgeResearch groups and wards have separate knowledge areas and apps.
- Released dataFor research, basebox connects only the datasets your hospital provides for it, for example pseudonymised from the data integration centre or from your data platform.
- Everything loggedEvery access, in the chat or via the interface, is recorded in the audit log with person and group.
Runs on your infrastructure.
basebox brings its own Kubernetes. If your hospital already runs a cluster, basebox runs there too.
- On existing hardware in a few daysCoordination with your IT, installation in one day, training.
- Sign-in via your identity providerRoles and groups come from your directory.
- Plan computing power togetherWhich GPUs serve models and which stay free for research computing.
- FHIR connectorReads clinical data via FHIR R4. Which profiles your hospital information system (HIS) or data integration centre uses (ISiK, the German core data set of the Medical Informatics Initiative), we clarify in advance and set up the connector accordingly.
- Data platformsIf your hospital runs a data platform based on openEHR or FHIR, for example from Better, vitagroup or i-engineers, basebox connects it, read-only by default.
For hospitals with critical infrastructure (KRITIS) obligations.
basebox provides the audit log, evidence and documents you need for B3S Krankenhaus, the German security standard for hospitals.
- Attack detectionOn request, you receive the audit log for your security monitoring (SIEM).
- Your server, your approvalNo vendor access without your approval.
- EvidenceISO 27001 certified, BSI C5 in progress.
- Supply chainSoftware bill of materials (SBOM) in full in your basebox, a defined process for reporting vulnerabilities.
- B3S KrankenhausMapping of our measures in progress, available on request.
basebox takes part in research.
TWIN-XEU Horizon Europe, 2026 to 2030Digital twins for personalised medicine, 19 partners, including the Technical University of Munich. basebox provides the platform on which the AI runs in the clinic.
InteRAGtBayVFPThe interactive patient record, with the German Heart Centre Munich.
OnkoBotZIMAn app for oncology: an AI assistant for oncologists, with the Technical University of Munich.“basebox on-premise lets us analyze patient data while maintaining confidentiality. It has transformed our team's collaboration while upholding privacy standards.”
Univ.-Prof. Dr. Keno Bressem, Medical Data Science in Oncology, University of Duisburg-EssenYou will find the price in your personal quote.
For hospitals with more than 2,000 people we calculate individually; universities and non-profit organisations receive 15% off.
Frequently asked questions
Yes. Any model with an OpenAI-compatible interface can be connected, following our checklist for runtime, memory, tool calls and German.
Yes. After a security audit, we add apps from start-ups or from your own staff to the App Store. Each runs as its own container on your server and uses basebox’s models, knowledge and connections with the same permissions as all other apps.
Not necessarily. basebox uses the computing power your IT allocates to it. Beforehand, we clarify together which GPUs are planned for models and which for research computing.
Through groups, knowledge areas and permissions from your directory. For research, basebox connects only datasets your hospital releases for it, for example pseudonymised from the data integration centre. Every access is recorded in the audit log.
Through our FHIR connector (FHIR R4). It is set up for each hospital, because hospital information systems and data integration centres implement FHIR differently: with their own profiles (ISiK, the core data set of the German Medical Informatics Initiative, vendor profiles), released data and sign-in methods. If your hospital runs a data platform based on openEHR, we connect it in the same way. The vendor provides access to the hospital information system; which data basebox reads, we agree with your hospital and the vendor.