You decide what your AI assistant gets to see
For medical practices, law firms and care services, a phone call is often about more than booking an appointment. voiceOne lets you decide, per AI agent, which personal data from the conversation is detected and processed at all.
You decide what gets detected
For each agent, you decide which of the following data should be detected. Every category can be switched individually — not just all or nothing.
Bank and ID data
Data that is clearly identifiable to a person and especially sensitive.
- IBAN
- Health insurance number
- Date of birth
- Credit card number
Contact data
Data through which a caller can be reached or located.
- Email address
- Phone number
- Postal code
- Street address
Off or Replace — you decide
Off
Default. Nothing changes in how the agent behaves.
DefaultReplace
Detected data is replaced with a placeholder before the text is sent to the language model. The caller themselves keeps hearing the normal, unaltered speech.
AvailableChecksum, not guesswork
Instead of just looking for number patterns, the agent checks whether a detected number could actually be a valid IBAN or health insurance number. This reduces false positives compared to plain pattern matching.
Data that is spelled out rather than read as digits is also detected — for example an IBAN dictated letter by letter, "D E eight nine three seven …".
Decide beforehand, not check afterwards
When health or client information comes up on a call, you don't want to check afterwards what ended up in the system. With voiceOne you decide that beforehand, per agent — not afterwards through a deletion request.
Frequently asked questions
Do you run a medical or therapy practice?
Discover mediOne — our specialised AI assistant for practices, with the same per-agent privacy settings.
Discover mediOneQuestions about setting this up for your agent?
We'll show you how to set the data types for your agent.