The NHS Ambient Voice Technology Self-Certified Supplier Registry was migrated to its new Digital NHS home in April 2026 and is described as a national capability to support safe and effective scaling of AVT across health and care. Digital Health reported on 28 April 2026 that G2 Speech had been added as the registry reached 23 suppliers.
The registry is a procurement and assurance tool — it helps NHS organisations identify suppliers who have self-certified against baseline requirements. But being on a registry is not the same as being clinically risk-free in every deployment context.
What the AVT Registry Is
A self-certified supplier registry intended to help NHS organisations with procurement and assurance. Suppliers must hold Class 1 Medical Device accreditation and current DTAC assessment. The registry provides a national list of suppliers who have demonstrated compliance with baseline safety, governance, and information-governance standards.
What Self-Certification Does and Does Not Mean
Self-certification means the supplier has attested to meeting the registry's requirements. It does not mean the MHRA has independently validated the clinical performance of the product. It does not mean local deployment is automatically safe — deployment requires local clinical safety review, DPIA, information governance assessment, workflow design, staff training, and ongoing monitoring.
The registry helps organise supplier evidence and provides a starting point for procurement. Local governance — including the clinical safety case, data processing agreements, consent processes, and user training — remains the responsibility of the deploying organisation.
What Clinicians Should Check Before Using an AI Scribe
Who reviews the note? The clinician must review and approve every AI-generated output before it enters the patient record. This is not optional — it is the governance expectation.
Is the patient informed? NHS England requires transparency — patients should be told at the start of the session that an ambient scribe is in use.
Where is data processed? UK GDPR requirements apply. Audio recordings should typically be deleted once a verified summary is produced.
What happens to audio? Retention beyond summary production requires documented justification.
Is the output coded or free text? SNOMED coding suggestions carry different clinical implications from narrative text — particularly regarding disease registers, QOF, and prescribing.
Are hallucinated statements possible? LLM summarisation can produce plausible but incorrect content. Clinician review must check for fabricated examination findings, false diagnostic certainty, and incorrect medication details.
Is there an audit trail? Tracking what the AI generated, what the clinician modified, and what was saved supports governance and medico-legal defensibility.
Can errors be reported? A feedback mechanism allows quality improvement from real-world use.
Where iatroX Fits
The rise of AVT makes clinical knowledge verification more important, not less. Once AI writes the note, clinicians still need to verify the clinical content. iatroX is designed for that adjacent professional workflow: source-grounded answers, provenance, calculators, CPD, and feedback mechanisms.
AI can help draft the note. iatroX helps professionals check the knowledge behind the decision.