Tocndix designed a speech-recognition medical assistant for a healthcare organization: the app transcribes the consultation, drafts the visit report and lists follow-up actions for the doctor and the patient. The doctor then reviews and signs off. A look back at a human-centered AI project.
The need: less time writing, more time for patients
After each consultation, the doctor has to write a report and note the follow-up steps. This paperwork takes time and can get pushed to the end of the day, when things are easily missed. Our client wanted to reduce that time.
The solution: an assistant that listens, drafts and suggests
- Transcription: the consultation is transcribed using speech recognition.
- Visit report: AI drafts a report from the transcript.
- Follow-up actions: the application lists follow-up actions for the doctor and for the patient.
- Human validation: the report and the actions are finalized only after the doctor has reviewed and approved them.
AI prepares, the doctor decides
In healthcare, keeping a human in the loop is non-negotiable: the assistant prepares the work, the doctor reviews and validates it. This kind of project also requires hosting, access controls and data handling suited to the specific requirements of health data. For a tool like this, we recommend testing with doctors on real cases before rolling it out widely.
What this project shows
A useful AI application does not need to replace the professional: it takes the repetitive work off their plate and leaves the decision to them. It’s the same approach we take in our other projects, including industrial ones: a specific process, AI connected to existing tools, and human validation at sensitive steps.
Explore our AI projects · Secure AI applications · Talk about your project
Frequently asked questions
Is the report sent automatically?
No. The doctor reviews and validates it. AI prepares; it does not decide.
Can this kind of assistant be adapted to other fields?
Yes: any job that produces a report after a conversation (site inspection, client meeting, audit) can use the same approach.