AI you can put in front of an inspector
We don't hide our AI behind marketing words. Every AI system in MyCareAudit is listed below in one honest register — what it does, what data it touches, and the human controls around it. Outputs are labelled, a person always decides, and your data is never used to train models.
Our AI governance principles
The rules every AI feature in the platform follows — no exceptions.
AI assists, humans decide
No AI feature in MyCareAudit makes an autonomous decision about a person’s care. AI proposes, summarises or drafts — a named person always reviews and decides. There is no automated decision-making with legal or similarly significant effect (UK GDPR Article 22).
Every AI output is labelled
Anything produced by a generative model — draft reports, summaries, suggested actions — is clearly marked as AI-generated in the interface, so staff never mistake a draft for a verified record.
Pseudonymisation by default
Where a feature sends care records to a model, direct identifiers are minimised or pseudonymised first. We process the minimum data needed for the task (data minimisation, UK GDPR Article 5).
Your data is never used to train models
Your organisation’s data is used only to serve your own requests. It is not used to train, fine-tune or improve any shared or third-party model.
UK data residency & security
Data is held in the UK under our existing security controls (encryption in transit and at rest, role-based access, full audit logging). AI features inherit the same controls as the rest of the platform.
You stay in control
AI-assistive features can be reviewed per organisation, and every AI interaction on your account is recorded in your in-app AI Usage Register so you can evidence oversight to inspectors.
Safeguards on every AI draft
How we keep a person in charge of every record the AI helps to write.
A full audit trail
Each AI draft records which model wrote it, which version of our instructions it followed, who approved or rejected it and which fields a person changed before saving. Managers can view and download this trail.
Topics the AI never fills in
DNACPR and resuscitation wishes, allergies, medicines and doses, safeguarding concerns and choking or swallowing risks are always left for a person to complete, even when they are mentioned in a recording.
Time-stamped sources
When a draft comes from a voice recording, each field shows the moment in the recording it came from and the words heard, so the reviewer can check it quickly.
Drafts until approved
Nothing the AI writes becomes part of a care record until a named person approves it. Gaps are marked for the reviewer to add rather than guessed.
Every AI system we run
A living inventory of the AI (and the deterministic, rules-based intelligence) inside MyCareAudit. Your customers and inspectors can see exactly what is automated and how it is controlled.
Last reviewed 2026-10-01
AI notes summary & before-your-visit prompts
Care ManagementGenerative AISensitivity: LimitedSummarises one person's recent daily care notes and open signals, and suggests 3-5 things for the next care worker to check on their visit. Every point links to the notes or signals it came from; points without a valid source are removed automatically. It does not give clinical instructions and nothing is written to the care record. AI-assisted summary for review, not a decision.
AI Voice Assessment (holistic assessments, care plans & risk assessments)
Care ManagementGenerative AISensitivity: LimitedTranscribes a consented recording of an assessment conversation and drafts holistic needs assessment sections, care plan or risk assessment content from what was said. Sections that were not discussed are left blank. In review mode it compares the conversation with the current care plan section or risk assessment and only suggests changes, each with a reason and quote, which the reviewer accepts or rejects one by one. Clinical guardrail: on high-risk topics (medicines, insulin, swallowing, pressure areas, seizures, oxygen, tubes, allergies, DNACPR) it only quotes what was said, with a time marker, and the reviewer is warned to check against the prescriber / SALT / nurse instructions. The draft is for review, not a decision — a manager or senior carer edits and signs it off.
Add via voice (form filling across care records)
Care ManagementGenerative AISensitivity: LimitedTranscribes consented speech and fills the fields of the care-record form that is open (daily notes, health and clinical records, goals, planning records). Fields not mentioned are left blank. Nothing is saved automatically: staff check every field and press Save themselves.
AI Audit Report Drafting
Audits & ComplianceGenerative AISensitivity: LimitedDrafts a narrative audit report and improvement suggestions from the evidence and scores a user has already entered.
Care Assistant (chat)
Care ManagementGenerative AISensitivity: LimitedAnswers compliance and “how-do-I” questions and helps staff find the right guidance inside the platform.
Care Record Insights
Care ManagementAssistive AISensitivity: Higher sensitivityHighlights themes and possible gaps across daily notes and care records to support review — never to replace clinical judgement.
Care Intelligence Signals
Care ManagementDeterministic (rules-based)Sensitivity: LimitedFlags deviations from each person’s personal baseline (e.g. weight, mood, activity) for a human to review.
Clinical Scoring Tools
Care ManagementDeterministic (rules-based)Sensitivity: LimitedCalculates recognised clinical scores (e.g. risk tools) from entered observations using published formulas.
Resident 360° Summary
Care ManagementDeterministic (rules-based)Sensitivity: LimitedAssembles a one-page “understand this person before care begins” snapshot from existing baseline, signals and care-plan status.
Regulatory Intelligence
Quality & GovernanceAssistive AISensitivity: MinimalMonitors published regulator sources and summarises what changed, for a human to interpret for their service.
Customers on the platform get a live version of this register scoped to their own account — including a record of AI activity on their organisation — so they can evidence AI oversight in their own CQC or Ofsted submissions.
Put responsible AI to work in your service
Download the AI Trust Pack to share with your board, DPO or inspector — or start a free trial and see the labelled, human-in-the-loop AI for yourself.