Ready to demonstrate
Explainable rules and exclusions, human override with reasons, pseudonymised existing-child compatibility, Ofsted evidence workflow, audit design and separated roles.
Clear answers about professional accountability, safeguarding, provider verification, data access and procurement readiness.
Explainable rules and exclusions, human override with reasons, pseudonymised existing-child compatibility, Ofsted evidence workflow, audit design and separated roles.
Council DPIA and equality assessment, production tenant enforcement, penetration and accessibility testing, contractual controls, incident/continuity plans and pilot validation.
Yes, within a carefully limited model. CareMatch first assures the provider organisation. A council administrator then approves that already-assured organisation to participate in the council’s own workspace.
Important: “Local authority admin” means workspace administrator, not global CareMatch administrator. Production enforcement must sit on the server and database, not only in the interface.
No. CareMatch ranks evidence against configurable rules and explains the result. An authorised Local Authority professional reviews the evidence and retains the final placement decision.
No. A triggered hard exclusion removes the home from the recommended shortlist. It remains visible in the audit record with the reason, but cannot be rescued by positive weighted factors.
No. Vacancy is only an availability check. CareMatch first assesses registration and Statement of Purpose fit, information quality, safeguarding, the impact on existing children, staffing, location, services and the care plan before any eligible home is ranked.
Suitable means every mandatory gate is recorded as met. Suitable with conditions requires named measures before admission. No-go means a mandatory requirement cannot be met. Insufficient information pauses the decision until critical evidence is obtained.
Yes. The authorised professional can override the ranking, but must record a reason. The recommendation, chosen home, decision-maker, time and rationale are retained together.
The proposed design compares pseudonymised risk, vulnerability, trigger and safeguard factors. Local Authority users see the compatibility conclusion and professional question, not another child’s identity or full case record.
The interface should label missing, self-declared and expired evidence, reduce confidence and request professional review. Absence of evidence must never be treated as evidence of safety.
This must be agreed in the contract and data-flow assessment. A Local Authority will commonly be controller for its referral information and CareMatch a processor, but the parties must document the actual purposes and responsibilities.
A DPIA should be completed before live child-level information is used because the service may involve vulnerable children, risk data, profiling and new technology. CareMatch should support the council’s assessment, not replace it.
No. Production access should be tenant-scoped and enforced on the server and database, with automated tests, logging and periodic access reviews.
No. Providers should see only their organisation’s records and referrals explicitly shared with them. Commercially sensitive evidence must not be exposed to another provider.
No. Platform administrators should manage tenancy, subscriptions and audit integrity without routine access to child-level or provider-confidential content. Any exceptional support access requires authorisation, purpose limitation and logging.
A production contract needs a documented retention schedule, deletion workflow, legal-hold process and support for access, correction, restriction and objection requests. Pilot data should be minimised and deleted or anonymised on an agreed date.
The provider supplies its legal identity, home name and URN. CareMatch staff compare them with the official Ofsted reports service, record the source URL, retrieval date and result, and schedule re-checks. Provider-entered ratings remain labelled unverified until that step is complete.
No. A provider may submit evidence but cannot change its status to independently verified. Verification is separated from the provider role and logged.
A production pilot should set an expiry window, require providers to reconfirm availability, show the last-confirmed time and automatically mark stale vacancies unavailable for recommendation.
No. Ofsted information is one assurance input. It does not replace child-specific safeguarding, existing-child compatibility, staff competency, therapeutic fit or professional judgement.
Yes. Each result should show the rules and weights used, hard exclusions, factor scores, strengths, concerns, existing-child compatibility and why it ranked above or below alternatives.
No. The current MVP demonstrates configurable, explainable rules. Outcome data is structured for future evaluation, but no claim is made that the system learns or improves itself.
At minimum: DPIA and equality assessment, information-sharing and processor terms, safeguarding review, threat modelling and penetration testing, accessibility testing, retention and incident procedures, service levels, business continuity and user acceptance testing.
Agree measures before launch: time to shortlist, suitable-response rate, professional agreement and override reasons, placement stability, breakdowns, false exclusions, data completeness, user experience and any differential outcomes.
A production service needs monitored availability, incident communications, recovery objectives, backups and a documented manual placement-search fallback. CareMatch must not become a single point of failure.
A living assurance pack should include architecture and data flows, role matrix, DPIA support, security test results, accessibility statement, sub-processor list, retention schedule, incident plan, matching methodology, change log and audit examples.
Read the ICO guidance on AI and data protection, its accountability and DPIA guidance, and the official Ofsted reports service.