The NHS Neighbourhood Health Framework, published on 17 March 2026, explicitly names AI-assisted triage as part of the digital infrastructure for neighbourhood health. The 10 Year Health Plan commits to expanding AI-assisted triage pilots, embedding online consultation tools through the NHS App, and making digital channels the default for patient access. The Medium Term Planning Framework goes further, describing AI triage and data-driven pathways as the mechanism for guiding patients to the right service.
This is not aspiration. It is policy with a timeline.
But there is an important distinction that Practice Managers and PCN Clinical Directors need to understand: what the framework requires and what it recommends are not the same thing. This post separates the two, explains what AI triage actually does in a primary care setting, and sets out why regulation matters more than functionality when evaluating which tools to trust.
What the Framework Says About Digital and AI
The Neighbourhood Health Framework describes a strong digital approach as critical to organising services around a person’s needs rather than organisational convenience. Three specific commitments relate directly to AI and digital triage.
AI-assisted triage pilots will be expanded nationally. The framework does not mandate that every practice adopt a specific AI triage tool. It commits to expanding pilots and making tools available. The direction of travel is clear: AI triage is where national policy is pointing. Whether your practice adopts it now or waits for local commissioning to require it is partly a matter of timing.
Online consultation tools will be embedded through the NHS App. From April 2026, the NHS must make at least 95% of appointments available after appropriate triage via the NHS App across all care settings. This creates a direct link between triage infrastructure and the NHS App, the government’s chosen front door to the NHS by 2028.
A digital-first approach is being made a formal commissioning expectation. The Medium Term Planning Framework names My NHS GP, a feature set within the NHS App, as the mechanism for using AI triage and data-driven pathways to guide patients to bookings. Modern service frameworks to be published will set out plans for digital-by-default care where clinically appropriate.
What the framework does not do is mandate a specific AI triage product. That is a local decision, subject to ICB commissioning intentions and GP contract requirements. But the framework creates strong strategic pressure to evaluate and adopt AI triage tools. Practices that do not have a structured digital access model are increasingly out of step with the direction of commissioning.
What AI Triage Actually Does
There is a lot of noise around AI in healthcare, and a lot of tools that use the label loosely. For the purpose of understanding what AI triage means in a primary care context, it is worth being precise.
AI triage in primary care is a clinical decision support tool. It presents a patient with a structured symptom questionnaire, applies a clinical algorithm to their responses, assigns an urgency level, and recommends a care pathway. The output is not a diagnosis. It is a structured assessment that helps clinicians and care coordinators make faster, more consistent routing decisions.
It captures presenting problems in structured form. This is different from a free-text online form or a basic e-consult submission. Structured symptom capture means every contact produces a consistent, coded record of what the patient presented with. That record is what enables MDT routing, population health analytics, and audit-ready compliance data.
It operates across both online and telephone channels. Patients who submit online go through the same questionnaire as patients whose presenting problem is entered by a member of staff from a phone call. The result is a single, consistent data set across all contact channels. This is what total triage means: not forcing patients online, but applying the same structured process regardless of how they contact the practice.
It routes to the right professional, not just the GP. The value of AI triage is not just speed. It is accuracy. When a patient presents with a musculoskeletal problem, the system routes to a physiotherapist. When a patient has a medication query, it routes to a pharmacist. When a patient presents with red flag symptoms, it flags for urgent GP review. This routing happens consistently, without depending on a receptionist’s judgement or a GP’s capacity to review every contact manually.
Why Regulation Matters More Than Features
The NHS Neighbourhood Health Framework’s expansion of AI-assisted triage has been welcomed by clinicians, but with an important caveat. One clinical advisor quoted in Digital Health following the framework’s publication said the focus must be on scaling tools that have undergone proper MHRA regulation, not simply tools that use AI as a marketing term.
That caution is well-placed.
A CE-marked medical device for clinical decision support has been through a conformity assessment demonstrating that it meets safety, performance, and quality standards for medical use. It is accountable to the MHRA. Its clinical logic can be audited. It carries a defined risk classification. If something goes wrong, there is a regulatory framework and a liability trail.
A patient-facing digital form that uses a language model to generate symptom guidance is not the same thing, even if it is marketed as AI. The clinical safety standards are different. The accountability is different. The regulatory status is different.
Klinik’s platform is a CE-marked Class IIa medical device, cleared for use as a clinical decision support tool in primary care. This matters for three reasons.
First, it means the clinical algorithm has been independently validated against safety standards. The urgency assessments the system generates are not best guesses. They are the output of a validated clinical logic engine.
Second, it means your practice has a defensible basis for the routing decisions the system makes. If a patient is triaged as non-urgent and subsequently deteriorates, the audit trail shows a documented, CE-marked clinical assessment, not a receptionist’s judgement call.
Third, it means the tool meets the regulatory expectations that the framework’s push towards AI adoption will increasingly require. As NHS England expands AI-assisted triage pilots and ICBs begin to assess which tools to recommend or commission, regulatory status will become a selection criterion, not a nice-to-have.
The NHS App and What It Means for Your Triage Setup
The government has committed to making the NHS App the default front door to the NHS by 2028. Online consultation tools will be embedded within the App. From April 2026, 95% of appointments must be available via the App after appropriate triage.
For practices, this creates a question about architecture. If a patient contacts your practice via the NHS App and is routed into an online triage flow, what happens next? The NHS App provides the front door. Your triage system is the decision engine behind it. The two need to be compatible.
Klinik’s platform is designed to integrate with existing NHS primary care infrastructure. It functions alongside EMIS and SystmOne, outputs structured data in formats clinical systems can receive, and is built for the multi-channel environment the NHS App creates, capturing contacts that arrive via the App alongside those that arrive by phone or direct online submission.
As NHS App triage integration develops over 2026 and 2027, practices that already have structured total triage in place will find integration straightforward. Practices starting from a basic online form or phone-only triage will face a more significant transition.
What the Data Shows
The framework’s push towards AI triage is backed by real-world evidence from practices that have already made the transition.
At Priory Medical Group in York, implementing structured AI triage via Klinik allowed the network to handle 45% more appointments without adding staff: 258,627 appointments in a year compared to 177,883 prior. The same evaluation calculated capacity-releasing savings of around £300,000 in a single year, drawn from shifting GP appointments to pharmacists, reducing NHS 111 out-of-hours usage, and cutting administrative task volume.
Across Klinik-using practices, 83% of staff reported that patient queries were being directed more quickly to the right place than before. Phone contact dropped from 99% of all contacts to 30% at one practice, with calls answered within 5 minutes rather than the previous 30-minute average.
These are not AI experiment outcomes. They are the operational reality of a practice running structured total triage at scale.
What Your Practice Should Be Evaluating Now
The framework does not set a deadline for AI triage adoption. But it sets a direction, and the commissioning environment is moving that way. Practices that are evaluating digital triage tools now should apply three tests.
Regulatory status. Is the tool a CE-marked medical device? If not, what is the clinical safety basis for its urgency assessments?
Channel coverage. Does the tool capture telephone contacts through the same structured process as online submissions? A tool that only works for patients who submit online will leave a gap in your data and your compliance record.
Data output. Does the tool generate structured, coded demand data that can feed into your clinical system, ICB reporting, and population health management? Or does it generate a form submission that someone still has to review manually?
A tool that passes all three tests is doing what the framework’s digital shift requires. One that fails any of them is building a partial solution that will need replacing as commissioning requirements tighten.
Frequently Asked Questions
Does the NHS Neighbourhood Health Framework require my practice to use AI triage?
No. The framework commits to expanding AI-assisted triage pilots and embedding online consultation through the NHS App, but it does not mandate a specific tool or require every practice to adopt AI triage by a set deadline. However, the direction of commissioning is clearly towards structured digital access, and practices without a coherent digital triage model are increasingly at risk of being out of step with local commissioning expectations.
What is the difference between AI triage and an online consultation form?
An online consultation form captures free-text patient input and sends it to a clinician for review. AI triage presents a structured symptom questionnaire, applies a validated clinical algorithm to the responses, assigns an urgency level, and generates a routing recommendation. The difference is consistency, structure, and clinical validation. AI triage produces coded data; an online form produces text.
Why does CE marking matter for AI triage tools?
CE marking as a medical device indicates the tool has been through a conformity assessment and meets safety, performance, and quality standards for clinical use. It means the clinical algorithm is accountable to the MHRA, the tool has a defined risk classification, and there is an auditable regulatory basis for the clinical decisions it supports. Tools marketed as AI that are not CE-marked medical devices have not met these standards.
What is My NHS GP in the NHS App?
My NHS GP is a feature set within the NHS App described in the Medium Term Planning Framework as the mechanism for using AI-assisted triage and data-driven pathways to guide patients to appointments. It is part of the government’s plan to make the NHS App the default front door to the NHS by 2028.
How does AI triage generate savings?
By accurately routing patients to the right professional at first contact, AI triage removes the need for a GP to review every contact before it can be allocated. Contacts that appropriately go to pharmacists, physios, or clerical staff are handled without consuming GP time. At Priory Medical Group, this shift generated around £143,000 in savings from pharmacist routing alone, with further savings from reduced NHS 111 usage and administrative task reduction.
Is Klinik’s platform a CE-marked medical device?
Yes. Klinik is a CE-marked Class IIa medical device cleared for use as a clinical decision support tool in primary care. It integrates with EMIS and SystmOne, operates across online and telephone channels, and generates structured demand data compatible with clinical system and ICB reporting requirements.