Executive summary
In 2025, the challenge of managing patient demand, shortening waiting lists, and ensuring patients are seen in the right place at the right time is a central priority for the NHS. AI-enabled triage has emerged as a powerful and proven solution to this challenge, with validated deployments across UK primary care, emergency departments, dermatology, and musculoskeletal (MSK) pathways showing it can safely redirect demand, surface clinical risk earlier, and improve operational efficiency.
A number of innovative tools are now live in UK clinical practice. Klinik is delivering workflow and cost benefits at a PCN level for GP online triage. eConsult eTriage is streamlining patient flow in emergency and urgent care centres. For cancer, C the Signs is helping GPs with risk stratification, while Skin Analytics' DERM has been conditionally recommended by NICE for dermatology triage. And in MSK, Flok Health is delivering a CQC-approved AI-led physiotherapy service. For any NHS organisation, the key to harnessing these benefits lies in a structured approach to procurement and deployment, underpinned by the national assurance frameworks of DTAC, DCB0129/0160, and, where relevant, the NICE Early Value Assessment (EVA) pathway.
Market snapshot: what “AI triage” means in 2025
"AI triage" is a broad term for a range of technologies designed to intelligently sort and direct patient needs. In 2025, this spans:
- Self-service patient tools for symptom assessment and navigation.
- "Total triage" systems in general practice that manage all inbound patient demand.
- Streaming tools in emergency departments to prioritise patients on arrival.
- Specialty-specific pathways that use AI to assess risk for conditions in dermatology or MSK.
The national direction of travel is clear: NHS England is actively exploring AI-enhanced triage for national services like the NHS App and 111, signalling a system-wide commitment to this technology.
UK-used tools at a glance
Primary care digital triage
- Klinik Healthcare Solutions: An AI-enabled online triage and patient-flow management system for general practice. UK case studies from PCNs report significant reductions in administrative tasks and demonstrable cost savings.
- Accurx Patient Triage: While historically more rules-based, Accurx is a widely used digital triage platform that integrates with the NHS App and provides a benchmark for workflow and efficiency evidence.
- eConsult (Primary Care): A long-standing online consultation provider that received an NHS AI Award to develop intelligent routing, helping to direct patient requests to the right team member automatically.
Emergency & urgent care
- eConsult eTriage: A dedicated module for emergency departments and urgent treatment centres that provides automated patient check-in and clinical prioritisation on arrival.
Cancer risk in GP
- C the Signs: An AI-powered decision support tool designed to help GPs identify patients at risk of cancer at the earliest possible stage. An independent UK evaluation has signalled an uplift in cancer detection rates, and the tool is reportedly in use across a large number of practices.
Dermatology
- Skin Analytics DERM: An AI tool that assesses images of skin lesions to help triage urgent suspected skin cancer referrals. It has a conditional recommendation from NICE for use in the NHS under the Early Value Assessment (EVA) pathway.
Musculoskeletal (MSK) / physiotherapy
- Flok Health: A CQC-registered provider that delivers an AI-operated physiotherapy triage and treatment service. Deployments with NHS partners in Lothian and Cambridgeshire have reported that the service has halved waiting times.
111 & navigation
- NHS 111 online pilots: NHS England is actively trialling the use of AI to support and enhance the triage and navigation functions within the national NHS 111 online service.
Evidence signals (what the data actually show)
- Primary care: PCN-level case studies with Klinik have reported greater than 20% reductions in administrative tasks and significant financial savings.
- Cancer triage: An independent evaluation of C the Signs by Health Innovation East showed a measurable improvement in cancer detection rates.
- Dermatology triage: The NICE EVA for DERM supports its time-limited use in the NHS, providing a clear pathway for evidence generation.
- MSK triage: NHS communications on the Flok Health deployments have cited significant reductions in waiting times for back pain and MSK services, as well as clinician hours saved.
Safety, regulation & assurance (UK specifics)
Safe adoption of AI triage tools in the NHS requires adherence to a clear governance framework:
- DTAC (Digital Technology Assessment Criteria): The mandatory baseline for procurement of any digital health tool.
- DCB0129/0160: The clinical safety standards. The supplier must provide their DCB0129 safety case, and the adopting organisation must complete a local DCB0160 safety case.
- NICE EVA: The pathway for emerging AI tools where the evidence is promising but still developing.
- MHRA Classification: Any tool that provides a diagnostic or risk-stratification output is likely to be a medical device and must have the appropriate UKCA/CE marking and MHRA registration.
How to choose (procurement & clinical criteria)
- Clinical fit: Does the tool match your specific pathway need (GP vs ED vs dermatology)? Does it have robust red-flag rules and clear escalation protocols?
- Operational fit: Does it integrate with your core EHR (EMIS, SystmOne, Cerner)? Does it connect with the NHS App?
- Measurement plan: Can you pre-agree the key performance indicators (KPIs) you will track, such as time-to-triage, referral appropriateness, and clinician time saved?
Implementation playbook (90 days)
- Baseline: For 2-4 weeks, measure your current demand, waiting times, and any relevant safety incidents for a single, well-defined pathway.
- Pilot: Implement your chosen tool using a stepped-wedge design. Ensure you have a clear "AI-off" fallback and a process for managing clinical disagreements with the AI's suggestions.
- Assure: Complete all your governance documentation (DTAC, DCBs, DPIA) and align with the NICE EVA framework if applicable.
- Publish: Share your outcomes—positive or negative—via AHSN or ICB channels to contribute to system-wide learning. Only plan a scale-up when your KPIs and safety thresholds are met.
Architecture Notes for Accuracy
When evaluating vendors, ask about their technical approach. Do they use provenance-first outputs with citations? Is the system a "black box," or does it use a more transparent architecture like Retrieval-Augmented Generation (RAG)? Crucially, are there non-negotiable human-in-the-loop controls for all high-risk decisions?
FAQs
- Are these tools actually used in the NHS?
- Yes. Examples include Klinik in GP PCNs, eConsult eTriage in EDs, C the Signs across approximately 1,400 practices, DERM under the NICE EVA pathway, and Flok Health in NHS Lothian and Cambridgeshire.
- Is AI triage safe to scale now?
- It can be, but only with proper assurance. This means passing the DTAC, having DCB0129/0160 safety cases in place, and, where indicated, following the evidence-generation requirements of a NICE evaluation.
- Does AI replace clinicians in triage?
- No. All current UK deployments emphasise augmentation, not replacement. The systems are designed to support human clinicians by providing better information and automating administrative tasks, with clear escalation pathways and audited decision trails.