Is Practice Management Software a Medical Device? (MDR)
Last updated: 25 October 2026
1. Regulatory classification
Ask Brigid's intended purpose is administrative practice-management. It is not intended or authorised for a medical purpose. DJG Media Limited, CRO No. 762838 does not hold or claim an EU MDR CE mark for the current product. The manufacturer recorded an administrative-purpose determination on 10 September 2026. It is the manufacturer's position, not an HPRA decision. The determination was made on 10 September 2026. The note assistant (declared on 28 September 2026) and the functions added after the determination are being assessed under Article 2(1) and MDCG 2019-11; the 10 September determination does not yet cover them. A changed feature set or a changed claim needs a fresh assessment. This page is not regulatory approval.
Software qualification is assessed under MDR Article 2(1) and the current MDCG 2019-11 revision 1 guidance. A qualification query to the Health Products Regulatory Authority was prepared but not submitted; no HPRA position is claimed. No declaration of conformity under any MDR class has been made. The accountable director declared an administrative intended purpose for each route enabled on 28 September 2026 (between 25 August and 28 September 2026); those declarations are the manufacturer's position, not an HPRA approval. Functions added later are listed in section 2 as not yet part of that written assessment.
- Provider: DJG Media Limited, CRO No. 762838, Coliemore House, Coliemore Road, Dalkey, Dublin, Ireland
- Product: Ask Brigid, in the Ask Brigid family with MyBrigid
- Current intended purpose: clinic administration, records workflow, booking, billing, task management, approved templates and staff-authored documents
- Current evidence: no notified body, Basic UDI-DI, UDI-DI or medical-device CE mark is held or claimed; the manufacturer's determination of 10 September 2026 is the qualification record
2. Intended use
Ask Brigid is intended for clinic staff to organise administrative work. It can:
- Store, retrieve, and organise clinic records and documents
- Support appointment scheduling, billing, reminders, and administrative task queues
- Transcribe clinician dictation and format clinician-supplied information
- Prepare administrative drafts for a person to review before use
- Support patient access, messaging, and consent administration
Administrative Generative AI is enabled only for functions with an administrative intended purpose. Functions (a) to (g) below each have a written intended-purpose declaration by the accountable director, and automated boundary tests pin their limits. Functions (h) to (l) were added afterwards and are being added to the written assessment. Staff are told when they are talking to an AI system, and AI-drafted output is labelled as AI-generated where a person reads it, with some parts still open (see section 4 and Compliance & Security). An authorised person must verify AI output before use. Patient consent cannot enable any route. Generative AI that makes a clinical judgement, and all patient-facing Generative AI, is disabled.
AI-assisted functions currently enabled. Functions (a) to (g) are each declared administrative: (a) a prose restatement of information already recorded in a patient's chart, shown to the treating clinician before a consultation and on request from the in-product assistant — it adds no opinion, ranking, risk grade or recommendation, and its qualification under MDR Article 2(1) is assessed against the manufacturer's determination described in section 1. The separate notes-enrichment route that writes into a patient's free-text record is disabled; (b) mailbox sorting: labelling and one-line summaries of the practice's own correspondence, putting each email in a queue order (Urgent, High, Medium, Low) for how soon staff should open it — an order for the mailbox, not a statement about a patient — staff to-dos raised from it, and reply drafts limited to logistics; (c) identifying which patient an emailed document belongs to and filing it, without interpreting the document clinically; (d) a setup assistant used during clinic onboarding that processes no patient data; (e) the note assistant, declared on 28 September 2026: it writes down what a clinician said, and tidies what they wrote, in a note or letter the clinician then reads and signs. It drafts from the consultation transcript, the clinician's own words and, when asked, what is already in the chart. Its instructions, held on the server where the user cannot edit them, forbid a diagnosis, differential, urgency or severity grading, triage, clinical coding, and any recommendation the clinician did not already make. The draft is labelled as drafted by AI and is not the record until the clinician signs it; (f) template field detection, which reads a blank form to find its fill-in fields, so that the practice's own paper can become a template it fills — it writes nothing but its own usage record, and the template is saved only after a person has seen the fields; (g) the patient-import identity read, which reads the identifying details on an imported document so that it is filed to the right chart, without interpreting the document clinically; (h) Ask Brigid Live, the voice mode of the assistant, which uses administrative tools only — reading patient records by voice is switched off for every practice; (i) tidying of dictated text, which corrects wording for the clinician to review; (j) the assistant's memory of staff preferences and open work; (k) suggestions of administrative jobs for staff to approve, where a practice switches them on; (l) lookup of SNOMED and ICD codes and the US NPI registry on request. Video visits are a call between clinician and patient and use no AI: nothing is recorded, transcribed or summarised. Disabled: symptom triage, differential diagnosis, result interpretation or urgency grading, risk scoring, dosing, medical-literature search, drug-interaction and drug-label lookup, recording of video visits, and all patient-facing Generative AI.
3. Excluded clinical purposes
Ask Brigid is not intended to diagnose, triage, monitor, predict prognosis, recommend treatment, interpret a clinical finding, calculate clinical risk, prescribe, or replace professional judgement. It must not be used in emergencies or as the basis for a diagnosis or treatment decision.
Do not use any Ask Brigid workflow as clinical decision support. If a workflow seems to give clinical advice, stop using it and tell us at support@askbrigid.com; it is not intended to.
4. EU AI Act transparency
EU AI Act Article 50 (transparency) has applied since 2 August 2026. Staff are told when they are talking to an AI system: the in-product assistant and the setup assistant identify themselves as AI, and Ask Brigid Live shows an AI disclosure before the microphone opens (Article 50(1)). AI-drafted notes, letters, summaries and email drafts carry a visible AI label where a person reads them before acting, and are marked as AI-generated in the stored record (Article 50(2) asks providers to mark AI output in a machine-readable way). These parts are not yet complete: AI-raised tasks are labelled when opened but not in task lists; the category and priority given to an email are not labelled; and machine-readable marking of exported documents and of emails the assistant sends is being added. No patient talks to an AI system in Ask Brigid or MyBrigid (see Compliance & Security).
In the manufacturer's assessment, Ask Brigid's AI functions are not intended for any use listed in Annex III of Regulation (EU) 2024/1689 and are not intended as a medical device or as a safety component of one, so they are not high-risk under Article 6. No AI Act conformity assessment is claimed. If that changes, we will update this page before the change is offered. The system inventory and provider/deployer role analysis are maintained as internal records. Article 4 (AI literacy) applies to us and to each clinic that uses the functions. Patient-facing Generative AI, and Generative AI that diagnoses, triages or recommends, remains disabled.
5. Human review and action controls
Every permitted administrative AI output must be checked by an authorised person before it is used. The platform applies confirmation boundaries to consequential actions:
- Drafts are not represented as sent, signed, prescribed, filed, or completed before the relevant server-confirmed action occurs.
- Role and permission checks still apply when Ask Brigid prepares or executes an administrative action.
- Users can stop, reject, or park work that needs review.
Administrative Generative AI runs only on routes with a declared administrative intended purpose and cannot be enabled by patient consent. Patient-facing Generative AI, and Generative AI that diagnoses, triages or recommends, remains disabled.
6. Audit, privacy, and change control
Each enabled administrative AI route carries a purpose classification, access control, production audit evidence, incident response and model-change records; any further route requires the same before it is enabled. Consent is not a substitute for an allowed purpose or lawful basis.
- AI activity records capture operational metadata such as feature, model, timing, and review state without creating a separate MDR technical-file retention claim.
- Platform and clinical audit logs follow the controller-approved, record-class retention schedule; no universal period is asserted.
- Material purpose changes are reviewed before release. A future medical purpose would require a fresh regulatory assessment before that use is offered.
7. Status updates
We will update this statement before any material change to the product's intended purpose or regulatory status. For qualification enquiries, contact legal@askbrigid.com.
8. Frequently asked questions
Is practice management software a medical device?
Not automatically. Under MDR Article 2(1) and MDCG 2019-11 (the European Commission guidance on qualification and classification of software), administrative software — scheduling, billing, records storage, templates, staff-authored documents — falls outside the medical-device definition when it does not process data to support a clinical decision. Software becomes a medical device when its intended purpose, as stated by the manufacturer, includes diagnosis, monitoring, prediction, prognosis or treatment.
What is Ask Brigid's intended purpose?
Administrative practice-management: clinic administration, records workflow, booking, billing, task management, approved templates and staff-authored documents, as declared in section 1 above. Ask Brigid does not diagnose, triage, monitor, predict prognosis, recommend treatment, interpret a clinical finding, calculate clinical risk or prescribe.
Does dictation or the administrative assistant change that?
No. Dictation transcribes speech to text for the clinician to review; where tidying is switched on, the AI only corrects wording and the clinician reviews the result. Neither analyses or interprets clinical content. Administrative Generative AI is limited to the functions in section 2 (drafting the note or letter the clinician recorded or dictated, prose restatement of information already in the chart, mailbox sorting, document filing, reading a blank form to find its fill-in fields, filing an imported document to the right chart, onboarding, voice mode of the assistant and code lookup) — none of it diagnoses, scores risk or recommends care.
Does Ask Brigid hold a CE mark?
No. No notified body, Basic UDI-DI, UDI-DI or medical-device CE mark is held or claimed. The manufacturer recorded its administrative-purpose determination on 10 September 2026, and a future change to intended purpose or feature set would require a fresh assessment before that use is offered.
Source guidance: MDCG 2019-11 Rev.1 — Guidance on Qualification and Classification of Software (European Commission).
Related documents
Clinician Terms of Service · Data Processing Agreement · AI transparency and governance · AI Features (patient privacy) · Acceptable Use Policy
For questions about this statement or the software's qualification, contact legal@askbrigid.com.
Download compliance documents