Digital Health: Recare develops AI agents for clinic administration

Berlin-based healthtech provider Recare aims to develop an AI agent as a "smart hub" for administrative processes. This is intended to relieve medical and nursing staff of documentation and coordination tasks in particular. To this end, the system automates, among other things, the creation and further processing of medical reports, handover protocols, and forms. Unstructured data from PDFs, scans, or free texts are to be extracted, structured, and converted into interoperable formats to make them usable across systems.

Recare has raised 37 million euros in a funding round, including an option for a further 7 million euros, for this project among others. The largest investor is the Norwegian inspection, insurance, and risk management group DNV, which thus becomes the largest shareholder of the company. CIBC Innovation Banking from Canada is also involved.

"Hospitals are under enormous pressure because medical professionals are spending more and more time on administrative tasks," says Recare CEO Maximilian Greschke. The AI agent takes over a large part of these activities and coordinates workflows across different systems. The rollout in Germany and other markets is now to be accelerated.

Software-as-a-Service

Recare operates a Software-as-a-Service platform for digital discharge management and aftercare to coordinate and automate care and other healthcare facilities. According to Recare, around two-thirds of German hospitals and more than 26,000 nursing staff work with the platform.

In addition to companies like T-Systems, the startup Voize also cooperates with Recare, which provides a language model for documentation in nursing care. The Recare AI agent currently has the Voize AI engine integrated modularly, Recare states upon request. Speech recognition and transcription run within the Recare Voice native app and are processed by the Voize AI model. Recare handles the structured further processing and integration into care and discharge processes – specifically, the direct integration of spoken text into any type of form.

The company emphasizes several security and data protection aspects: AI inference is performed client-specifically within the respective clinic context; data is not mixed. According to Recare, customer data is not used for training or improving AI models. Furthermore, a human-in-the-loop architecture is planned, where all AI results must be professionally reviewed and approved.

Technically, Recare relies on open standards such as HL7v2 and FHIR for integration into existing hospital IT, according to its own statements. Operation is as "Enterprise-Ready AI" with C5 Type 2 certification, including encryption, role-based access control, and audit trails, to meet GDPR requirements in healthcare.

(mack)