Medical Speech AI Platform: Corti Gears Up for Psychiatry and More

The medical speech recognition of the Danish company Corti is now also optimized for use in psychiatry. This is intended to enable reliable capture and analysis of psychiatric conversations with clinical understanding and context-sensitive interpretation. Especially in psychiatry, every phrasing, pause, or emphasis can be diagnostically significant, with frequent topic changes and incomplete narratives.

Central to this is Corti's clinical reasoning engine, FactsR, which runs continuously during consultations and recognizes medically relevant content in real-time. This automatically creates structured, traceable documentation that can be immediately transferred to patient records or clinical administration systems. In initial pilot projects, the system has already proven itself in the psychiatric field, according to Corti.

Shortly before, the company received, among other things, the BSI C5 certification for secure cloud applications, thus meeting international requirements for data protection, availability, and confidentiality. Furthermore, Corti's AI documentation assistant, “Corti Assistant MD,” is considered a Class I medical device, the lowest-risk class.

API Access for Developers

Via APIs, Corti enables access to various AI modules as needed, such as automatic speech recognition, text analysis, or medical coding. “The underlying models have been trained with millions of hours of documented doctor-patient conversations, including psychiatric sessions,” says Corti. Developers of health applications can therefore integrate AI functions into their systems without having to train models themselves.

In addition to a usage-based pricing model, where customers only pay for actual computing power—such as audio minutes or text tokens—the company also supports flexible deployment models, including enterprise-ready infrastructures.

The Corti Assistant has also been available as a mobile app for iOS and Android devices, including the Apple Watch, since May 2025, allowing the use of central speech recognition and documentation functions in clinical practice.

The original goal was to support fire departments and emergency services in making faster decisions. Today, the company specializes in AI for medical speech and text processing, develops its own models, and offers them via an API. Partners like Dedalus, Medatixx, and Philips Speech Processing have integrated the offering into their systems.

Comparable initiatives are also emerging in Germany: The University Medical Center Hamburg-Eppendorf (UKE) is currently offering its solution for AI-supported medical speech processing with Orpheus—and thus also offers a European alternative to the offerings of large tech companies.

(mack)