Frequently asked questions

No. HASTE-3D is in development. It is not in clinical use, and it is not available to patients or to NHS organisations. This page describes how we are designing it and what we intend to do. It does not describe a service in operation.

HASTE-3D (Holographic AI Smart Triage Engine - 3D) is being developed as an AI-powered remote triage system designed for NHS urgent care and A&E settings. It is intended to combine AI-supported decision support with the remote presence of a clinician, to help people reach the right care and to help clinicians identify urgent need quickly and safely.

Holographic refers to a three-dimensional remote presence of a clinician. AI refers to the decision support that helps analyse the information provided. Smart Triage Engine refers to the system that helps prioritise how urgently someone needs care and where they should receive it.

Triage is the process of deciding how urgently someone needs care and where they should receive it, for example self-care, a GP, an urgent treatment centre or A&E.

No. HASTE-3D is designed to work alongside existing NHS pathways, including NHS 111, and not to replace them. Clinicians remain responsible for triage outcomes.

It is designed for NHS organisations and the clinicians who run urgent care and A&E triage, and for the patients they serve. The patient groups and settings are set out in a written intended purpose, which governs our safety and regulatory work.

Urgent care and A&E services are under pressure. Some people wait for care that another route could provide sooner, and others who need urgent attention must be identified quickly. HASTE-3D aims to support clinicians in making those decisions safely.

HASTE-3D is a subsidiary of Amaizonia ltd and is led by its founder and Chief Product Officer. We welcome input from clinicians, researchers and technical specialists.

HASTE-3D is being designed to work in outline like this: information about a patient’s symptoms is gathered; the AI suggests how urgent the situation may be and what care route may fit, with its reasoning and level of uncertainty; a clinician, present remotely, reviews this, speaks with the patient and decides the outcome. The exact steps would depend on the NHS setting where it might one day be used.

It refers to a three-dimensional remote presence of a clinician. The intention is to support clear communication during remote consultations. It is a way of connecting people, not a replacement for clinical judgement.

The AI is designed to support triage by analysing the information provided and suggesting an urgency level and a suggested pathway, explaining its reasoning and flagging where it is uncertain. It is not designed to make final decisions.

In the design, the clinician reviews the AI’s recommendation, speaks with the patient, can override the AI at any point, and owns the outcome.

No. HASTE-3D is being designed so that every triage outcome with clinical consequence is owned by an identifiable clinician.

HASTE-3D is being designed to escalate. If the system is uncertain or a case is outside its scope, the case would be handed to a clinician. A separate rules-based safety layer is planned to watch for defined warning signs and direct people to emergency care.

That is a design requirement. Patients would be told when AI is involved and what it does.

Yes, that is a design requirement: patients would be able to ask for a human at any time.

We are designing HASTE-3D so that a non-digital route to care remains available, and so that nobody is disadvantaged for preferring not to use it.

Accessibility is a design goal, including for people with disabilities, limited digital access and different language needs. We plan to test with a range of users and to assess performance across different groups. We welcome feedback.

Guardrails are technical and human controls that keep an AI system within safe limits: a defined scope, rules it cannot override, limits on what it can output, and human oversight.

Our approach is to build layered guardrails into the design: checks on what goes in, controlled model behaviour, validation of what comes out, clinician review, and monitoring once any future use begins. No single control is to be relied upon, so a failure at one layer can be caught by another.

It is a planned rules-based layer, separate from the language model, that screens for defined red-flag presentations and escalates them. The design requires that the AI cannot switch it off or talk its way around it.

Language models can produce incorrect or invented content. Our design approach to reducing this risk includes grounding responses in approved clinical protocols, restricting output formats, validating outputs against safety rules, and requiring clinician review. We plan to test for it at every stage.

We intend to work to the NHS clinical risk management standard DCB0129 for manufacturers, and to support organisations with DCB0160. This would include a hazard log, a clinical safety case, and oversight by a Clinical Safety Officer, who must be a registered clinician. Naming a standard does not mean we hold certification against it.

Our plan is to use assurance gates: simulation, retrospective evaluation on appropriately governed data, shadow mode alongside live clinical triage, supervised use, and then wider use. Progress is based on evidence, not dates, and a gate is passed only when its evidence has been reviewed and signed off. No patient would be exposed to the system before the earlier gates are met.

We plan to assess performance across age, sex, ethnicity, language, disability and digital access, to test with diverse cases, and to keep a non-digital route to care available.

Our design includes tracking of performance, drift and incidents. Any change to the model, prompts, data or scope would go through documented change control and re-testing before release.

Our incident process is designed to record every incident, including near misses, contain the issue, and carry out a clinical review of any patient impact. We would report as required, including to the MHRA for device-related incidents, to the ICO within 72 hours for notifiable personal data breaches, and to affected NHS organisations. We would then find the root cause, correct it and re-test before the system returned to use.

Under our framework, clinicians would be accountable for triage outcomes, and HASTE-3D’s leadership for the safety framework. The Clinical Safety Officer role is designed to be able to stop progress at any gate.

Yes. We have an overview of our approach, which we share on request. Please contact us using the details at the end of this page.

It is being designed to use only the information needed for triage, such as symptoms and relevant history, provided by the patient or recorded by clinicians. We aim to collect only what is necessary.

Yes. UK GDPR and the Data Protection Act 2018 apply to personal data, and health data is special category data. Before any processing of patient data, we would identify a lawful basis, complete data protection impact assessments, and build in safeguards around automated decisions, including human review and the ability to contest outcomes.

The design principle is that data is held in controlled environments that meet NHS data governance requirements, with role-based, encrypted and logged access, so that only people who need it can see it.

Our design principle is that patient data is not used to train third-party models. Any use of data to improve the system would need a lawful basis, the appropriate approvals and safeguards, such as de-identification where appropriate.

No. HASTE-3D will not sell personal data.

Retention would follow NHS records and data protection requirements, and be set out in the privacy information for each setting where the system might be used.

Under UK data protection law, people can ask for access to their data, ask for corrections, and in some cases ask for deletion, restriction or to object to processing. They can ask for human review of an automated decision. You can raise a concern with the organisation holding your data, with us if we hold it, or with the Information Commissioner’s Office.

Our design includes encryption, role-based access, logging and regular security testing, including adversarial testing of the language model against misuse. We are working towards NHS data security expectations, including the Data Security and Protection Toolkit and Cyber Essentials.

Yes. Please see the cookie banner and our privacy notice for details and to manage your preferences.

We use it only to respond to your message, and handle it under our privacy notice.

Software that supports triage or clinical decisions can be regulated as a medical device in the UK, depending on its intended purpose. HASTE-3D is in development and has not been placed on the market. We are working to understand and meet the relevant requirements, including MHRA guidance.

No. HASTE-3D is in development and has not been approved, certified or clinically validated. We will only describe approvals, certifications or validation results once they have been formally obtained and can be evidenced, and we will publish them on this website.

We intend to work to the UK Medical Devices Regulations 2002 and MHRA guidance, NHS clinical risk standards DCB0129 and DCB0160, ISO 14971, IEC 62304 and ISO 13485, the NHS Digital Technology Assessment Criteria, the NICE Evidence Standards Framework, UK GDPR, the Data Security and Protection Toolkit, and the Equality Act 2010. Naming a standard does not imply certification.

Any study involving patients or NHS data needs the required ethical, regulatory and data approvals first. Participants will be told about the study and their choices.

UK regulation of AI in healthcare is changing quickly. We intend to review our framework regularly and to update this page when our approach changes.

We will share news and any formal milestones on this website.

No. In a life-threatening emergency, call 999. For urgent health advice that is not life-threatening, use NHS 111.

No. This website gives general information about HASTE-3D. If you are worried about your health, speak to a healthcare professional.

HASTE-3D is designed to be provided through NHS services, and is not intended to be paid for by patients.

Patients will be told how the system is used and what choices they have. Consent and information requirements will follow the law and NHS standards for each setting.

The groups for whom HASTE-3D is intended are defined in our intended purpose. Children and people who lack capacity need special consideration, which is part of our clinical safety work.

Contact us using the details below. If your care was provided by an NHS organisation, you can also use its complaints or patient advice service. For data protection concerns, you can contact the Information Commissioner’s Office.

Patient and public involvement is part of how we approach transparency, accessibility and acceptability. We welcome your views, so please get in touch.

HASTE-3D is not available for clinical use. We welcome conversations about collaboration and evaluation, and about possible future pilots once the earlier assurance gates are met. Any use with patients would follow our assurance gates and your organisation’s own governance.

Clinical input on triage pathways and escalation, advice on governance and data, and agreement on how the system should be evaluated.

As part of development we intend to prepare a hazard log and clinical safety case under DCB0129, and to support your organisation’s DCB0160 obligations. We would share documentation with partners as it is developed, under a confidentiality agreement where needed.

Interoperability, security and usability are part of NHS technology assessment. We will work with partners on the requirements for each setting.

Training is planned to cover how to use the system, how to read its reasoning and uncertainty, its limits, and how to recognise and avoid over-reliance on AI recommendations.

Yes. Clinical input is essential. Please contact us. Any advisory role is agreed in writing, with conflicts of interest declared, and advisers are free to raise safety concerns at any time.

We welcome collaboration on evaluation, AI safety, human factors and clinical validation. Collaborations are set out in formal agreements, with a confidentiality agreement where needed.

Please contact us. We share detailed material under a confidentiality agreement where appropriate.

We would assess partners on security, control over data flows, openness and fit with NHS requirements, and we favour approaches that keep data in controlled environments.

The public information on this website is an overview. We share detailed technical information only under a confidentiality agreement.

Please use the contact details at the end of this page.

HASTE-3D is a subsidiary of Amaizonia Ltd, a UK limited company.

Advisers and collaborators declare other roles, particularly with other triage or clinical decision support products. They can raise a clinical safety concern at any time, and we do not pressure anyone to sign off safety documents.

HASTE-3D is developing a Health Advisory Board of clinicians, researchers and patient and public representatives to challenge our approach. It is supported by Terms of Reference, a Code of Conduct and the Nolan Principles of public life.

Email Contact Us, or use the contact form on this website. Please do not send personal health information by email or through the contact form.

We review it regularly, and whenever our approach or status changes. The date of the last update is shown at the top of the page.