This article outlines the various AI initiatives at Konfidens. AI is rapidly becoming an integral part of several key areas in our product offering. Whereas many startups currently develop stand-alone AI solutions for a single use-case (such as generating clinical notes), we are building one of the first natively AI-powered practice management platforms and therapy marketplace. This has several key advantages:

Currently, we deploy AI in three main use cases:

AI-generated clinical notes

Currently working on a fully integrated feature to magically draft clinical notes for therapists using AI. Several startups have created stand-alone solution for this, but we are on track to become among the first practice management platforms to fully integrate this feature into the therapists existing workflow. Konfidens already automates almost all other administrative tasks for the therapists. With AI-generated notes, therapists using Konfidens can benefit from completely friction-free administration. The feature includes a dedicated mobile app for live recording and transcribing of the therapy session. The transcript is then processed using an LLM and a comprehensive set of instructions develop by our professional psychologists.

Freud – the therapy co-pilot

We’re piloting an AI co-pilot with access to a library of scientific books on mental health therapy. Opposed to ChatGPT and other publicly available LLMs, Freud is designed to strictly adhere to validated literature, cite sources for transparency and to empower the therapist to dive deeper into the underlying books and papers if need be. The copilot is built with a combination of langchain, probability calculation and the OpenAI LLM.

Working as a therapist is a lonely profession. You spend every day alone with your clients, without being able to discuss the challenges or issues with fellow practitioners, except for the occasional session with your supervisor. Freud is designed to be an always-available trusted partner and digital supervisor for the therapist – helping them discuss professional matters at their own convenience.

Client-therapist matching using machine learning

For our marketplace, we are building a state-of-the-art matching engine that leverages real-time access to actual treatment data from the underlying EHR. Our unparalleled access to data, makes this the most sophisticated matching ever built. The engineering is based on scientific research and holds the promise to reduce recovery time by an astonishing 40%.

In short, this will be achieved by screening new patients using a proprietary psychometric test. This data is then analysed and compared to a demographic/personality screening of each therapist on the platform in conjunction with longitudinal treatment progress data.

We have since inception started to develop the groundwork for this functionality, by building out the necessary infrastructure to gather and process the necessary data. First live AI-matching tests is planned in Q4 2025.

<aside> 📄

A more in-depth white paper can be found here:

White paper: AI Client-therapist matching

</aside>