Healthcare Communications DACH — Governed Voice AI for Patient Contact
Inbound patient contact — appointment booking, recall, triage hand-off — built for the data-sensitivity and scope limits of healthcare. PHI minimisation, scope-of-practice enforcement, Article 50 disclosure, and EU data residency are built in. EHR write-back is on the roadmap.
Integration stack
PHI minimisation
The agent collects only what the call needs. Patent-pending PHI compartmentalisation keeps sensitive turns out of general retention.
Scope-of-practice
No medical advice. Patent-pending scope rules route clinical questions to “a clinician will call you back” instead of an answer.
Article 50 + EU residency
Per-language AI disclosure in every call, with recordings handled under EU data residency — live.
EHR / practice-management write-back
Write-back into the systems healthcare teams already run is on the roadmap — not live today.
Why healthcare specifically
Patient contact is special-category data under GDPR, and a voice agent that drifts into clinical advice is a real problem. Healthcare teams also work under medical-device awareness (MDR), strict data-minimisation expectations, and a duty of care toward vulnerable callers. Generic voice AI handles none of this out of the box.
Audact builds these in as runtime primitives. PHI compartmentalisation (patent-pending) keeps sensitive data out of general retention. Scope-of-practice enforcement (patent-pending) stops the agent from giving a diagnosis or medication guidance — it books the patient and tells them a clinician will call back. Article 50 disclosure runs on every call, and vulnerable-person sensitivity adapts the interaction when a caller shows signs of distress or confusion.
Write-back into practice-management and EHR systems is on the roadmap — designed to close the“manual transcription”gap, but not live today. We are honest about what ships when.
Product Liability Directive · from 9 December 2026
If you white-label an AI agent to healthcare clients, the responsibility lands on you.
From 9 December 2026, the EU Product Liability Directive (2024/2853) can treat an agency that puts its own brand on an AI service as the “manufacturer.” In a personal-injury vertical like healthcare, that means strict (no-fault) responsibility can attach to your agency for harm a defective product causes. You become the party that has to show it acted responsibly.
A per-call, tamper-evident record of exactly what the agent did and did not say is the kind of documentation that helps you demonstrate due diligence — for example, that the agent stayed inside its scope, disclosed it was AI, and routed a clinical question to a human. Audact produces that record on every call, so the proof exists before you ever need it.
This is general information, not legal advice. Talk to your own counsel about how the PLD applies to you.Why your provider should carry the evidence →