NHS DSP Toolkit
The NHS Data Security and Protection Toolkit requires every organisation that handles NHS patient data to evidence, each year, that it meets the National Data Guardian standards for data security and protection. Mickai keeps NHS patient data on hardware the organisation owns, fully offline, so it is never transmitted to a cloud AI service and remains inside the organisation's own perimeter as sole controller. Because the data physically cannot leave the building, the toolkit's expectations around personal-data handling, access control, audit trails, secure configuration and incident response are enforced by architecture and evidenced through a post-quantum Open Audit Record that maps directly to the standards.
The DSP Toolkit assesses an organisation against the National Data Guardian standards, which cover personal confidential data handling, staff responsibilities, secure processes, managing data access, keeping systems protected, responding to incidents and holding data processors to account. NHS patient data is special-category personal data under UK GDPR and among the most sensitive an organisation can hold, and routing it through a cloud AI service creates a data-processor relationship to police, an egress path to defend and an assertion the organisation must make about an environment it cannot fully inspect. Access to patient data must be controlled and monitored, systems must be kept secure and current, incidents must be detected and reported, and any data processor must be assessed and evidenced. Each of these is harder to satisfy the more the organisation depends on a shared cloud. Mickai keeps NHS patient data, the models and the audit trail on hardware the organisation owns, fully offline, so there is no data processor to hold to account and no egress path to defend. Access is controlled and monitored locally through hardware-attested identity, every access is sealed to a tamper-evident Open Audit Record that evidences the audit-trail standard, and incidents are detected and contained within the organisation's own perimeter.
The 6 obligations this framework imposes, each met by construction on hardware you own and mapped to the subsystem that enforces it.
Sole Controller of Patient Data, No Cloud Processor
Because the sovereign brains process NHS patient data on hardware you own, no patient data is handed to an external AI service, so there is no data-processor relationship to assess, contract or police under the toolkit's data-processor standard. The organisation remains the sole controller of every record, and no shared model is exposed to patient data. This removes the processor from the chain entirely. Custody is enforced by architecture rather than by a processor agreement.
Managing Access to Patient Data
Mickai gates every access to NHS patient data behind hardware-attested identity and least-privilege clearance, satisfying the toolkit's expectation that access be controlled, role-appropriate and monitored. Access descends to the record level and can be scoped per role, so staff see only what their role permits. Nothing depends on a shared cloud identity layer. Access decisions are made and enforced inside the organisation's own perimeter.
Audit Trail Over Patient-Data Access
The toolkit expects a clear audit trail of access to personal confidential data, and Mickai meets this by sealing every access, use and disclosure of patient data to a post-quantum Open Audit Record. The trail captures the actor, the record touched, the purpose and the time, and it is tamper-evident and reproducible for an assessor or an information-governance review. There is no reliance on a vendor's logging tenancy. The audit trail is generated as a by-product of the work itself.
Keeping Systems Secure and Current
Mickai supports the standard on keeping systems protected by loading only signed, verified models and binaries and sealing every configuration change to the audit record on owned hardware. An unsigned or tampered artefact fails closed, and the organisation controls versions and refuses silent upgrades. There is no vendor changing the system outside the organisation's control. Secure configuration is enforced and evidenced locally.
Minimising and Governing Patient Data
Mickai processes only the patient data required for each clinical or administrative task and enforces retention against your own information-governance policy, all on owned hardware, supporting the toolkit's personal-confidential-data handling standards. There are no shadow copies in a cloud tenancy the organisation cannot inspect. Retention and minimisation rules are expressed as signed policy and applied before processing. The full data lifecycle stays inside the organisation's control.
Responding to Data-Security Incidents
Because the system runs inside the organisation's own perimeter, data-security incidents are detected and contained locally rather than inside a vendor environment, which supports the toolkit's incident-response standard and NHS reporting expectations. The sealed audit trail evidences exactly what patient data was accessed, when and with what effect, so response and reporting are accurate and first-hand. There is no dependency on a provider's disclosure. Containment and evidence are both local and provable.
The advantages hold across every framework, and they are architectural, not promotional. The third-party cloud-exposure vector is removed; your own physical, insider and compliance controls remain yours.
Scope inference, extraction, retrieval and storage on infrastructure you control. Document and test any external connector, support channel or transfer before describing a workflow as zero-egress.
Workflows with local models, data and supporting services can continue offline. Live external email, calendars and other services need an approved connection or controlled import, with clear unavailable and pending states.
Map prompts, documents, indexes, logs and backups to their processing and storage locations. Local hosting supports control of location; access, retention and any transfers still need review.
Prepare and classify source records, preserve permissions and evaluate answers against the originals. Retrieval and model training are separate choices; agree data use and licences for each.
Compare hardware and licences alongside integration, power, administration, support, updates, evaluation and recovery. Measure cost at the required workload and quality before claiming savings.
Define who holds keys, who can administer the system and which actions need approval. Test denied requests, revoked access and audit records. Physical security, insider risk and incident response remain operational responsibilities.
Agree version selection, signed update procedures, rollback, data export and licence rights. Reassess requirements as software and obligations change; owning hardware does not remove maintenance or compliance work.
How does on-premise AI help meet the NHS DSP Toolkit?
Mickai keeps NHS patient data on hardware the organisation owns, fully offline, so it is never transmitted to a cloud AI service and remains inside the organisation's own perimeter as sole controller. Because the data physically cannot leave the building, the toolkit's expectations around personal-data handling, access control, audit trails, secure configuration and incident response are enforced by architecture and evidenced through a post-quantum Open Audit Record that maps directly to the National Data Guardian standards.
Is there a data processor to hold to account when using Mickai?
No. Because the sovereign brains process NHS patient data on your own hardware, no patient data is handed to an external AI service, so there is no data-processor relationship to assess, contract or police under the toolkit's data-processor standard. The organisation remains the sole controller of every record, and no shared model is exposed to patient data. The processor is removed from the chain entirely.
How is access to patient data controlled and monitored?
Mickai gates every access to NHS patient data behind hardware-attested identity and least-privilege clearance, with access descending to the record level and scoped per role, so staff see only what their role permits. Nothing depends on a shared cloud identity layer, and every access is sealed to a tamper-evident Open Audit Record, which satisfies both the access-management and audit-trail standards.
Does Mickai provide the audit trail the toolkit expects?
Yes. Every access, use and disclosure of patient data is sealed to a post-quantum Open Audit Record with the actor, the record touched, the purpose and the time. The trail is tamper-evident and reproducible for an assessor or an information-governance review, held on hardware you own, and generated as a by-product of the work with no reliance on a vendor's logging tenancy.
How does Mickai support incident response under the toolkit?
Because the system runs inside the organisation's own perimeter, data-security incidents are detected and contained locally rather than inside a vendor environment, which supports the toolkit's incident-response standard and NHS reporting expectations. The sealed audit trail evidences exactly what patient data was accessed, when and with what effect, so response and reporting are accurate and first-hand.
Is Mickai a cloud service for NHS organisations?
No. Mickai is a Sovereign Intelligence Operating System that runs entirely on hardware the organisation owns, on-premise and offline, acquired as an owned asset rather than a metered subscription. The public cloud remains useful for non-regulated work; Mickai is the answer for the NHS patient-data boundary where special-category data cannot safely sit in a shared environment.
Bring NHS DSP Toolkit in-house.
Briefings are for organisations weighing a sovereign, on-premise deployment. Tell us about your estate and we will walk the obligations, the regulatory crosswalk and the deployment that fits.