Why patient-facing AI matters now
The "digital front door" for healthcare is no longer a futuristic concept; it's the 2025 reality. Patients are now routinely using sophisticated AI tools before they even think about booking a GP appointment. This trend is led by established global players like Ada Health, which has enterprise partnerships with payers and providers across multiple countries to help reduce the burden on acute services.
This shift is now being mirrored by NHS policy itself. The national 10-year digital plan includes the NHS “My Companion” AI for the NHS App, and local systems like NHS 111 Wales are already using multilingual AI assistants to guide patients. However, the capabilities of these tools are evolving far faster than the evidence. Studies, including those in JMIR, continue to show wide variation in the accuracy of symptom checkers. This makes it critical for UK clinicians and healthcare leaders to understand the landscape, know what their patients are using, and help signpost them to tools that are safe, effective, and aligned with UK practice, rather than tools that simply provide an unverified diagnosis.
Symptom & triage engines (global “front door” tools)
These are the most common tools patients encounter, designed to take a list of symptoms and provide a list of potential conditions or a triage recommendation.
- Ada Health: A well-established, CE-marked conversational AI symptom checker, widely used in Europe, the US, and Africa. It uses a conversational intake to produce a structured report and is often integrated at an enterprise level with health systems and insurers.
- K Health: A US-led platform that combines a free AI symptom checker with an integrated, low-cost (paid) clinician chat service. Its AI is trained on a large dataset of anonymised visit records.
- Buoy Health: A US-based decision support tool that helps users narrow down their symptoms and routes them to the appropriate level of care, from self-treatment to urgent care.
- Healthily / Dot™: A UK-born self-care and symptom checker, notable for its medically verified content library and AI-powered navigation.
It is crucial to remember that studies show the accuracy of these tools varies significantly. They are best used to help a patient inform their decision to seek care, not to diagnose their condition.
New category: patient-facing AI scribes & “medical-memory” apps
A new category of tool is emerging that focuses on the consultation itself, not just the triage before it.
- Aide Health “Mirror” (UK): Launched in late 2025, Mirror by Aide Health is one of the first AI scribes designed for the patient, not the clinician. With full consent, it listens to the consultation, auto-summarises the key points in plain English, and then makes that summary available for the patient to query later. This is designed to solve the well-documented problem of patients forgetting up to 80% of the information given in an appointment, aiming to improve recall and adherence.
- Advoca Health (UK): This platform shares a similar goal: building a patient-centred, longitudinal health record. It empowers patients to capture, organise, and make sense of their fragmented health information in a single, actionable app.
Virtual companions inside public systems
Public health systems are now building their own trusted AI assistants.
- NHS “My Companion”: Part of the UK's 10-year digital plan, this will be the official AI assistant embedded within the NHS App. Described as a "doctor in your pocket," it is intended to help patients navigate symptoms, manage medications, and find the right NHS service.
- NHS 111 Wales multilingual AI assistant: A live example of this in practice. This web-embedded tool pulls its answers directly from the trusted NHS 111 content library to guide patients.
Condition-journey & engagement bots
Beyond one-off triage, another class of AI tools is designed to support patients with long-term conditions.
- MayaMD: This platform combines an AI assistant and triage with remote patient monitoring (RPM) features, designed to keep patients engaged in their specific care pathways.
- Chronic condition coaches: Other platforms, like K Health, are expanding into specific modules for mental health or chronic disease management. In the UK, we are also seeing pilots of AI-powered physiotherapy services (like Flok Health) and discharge-support tools.
Accuracy, trust and regulation
The central challenge for all patient-facing AI is trust. Accuracy in symptom checkers varies markedly, and multiple studies have shown that physicians still outperform the best apps. For UK patients, the safest tools will always be those tied to NHS-approved content (like the 111 Wales assistant) or those with clear medical oversight.
For developers, as soon as a tool's advice moves from generic information to a specific treatment recommendation or diagnosis, it may be classified as a medical device and must meet MHRA regulations and the NHS DTAC procurement standards.
Where clinician-grade tools cross over (the iatroX angle)
The iatroX platform is built for clinicians on a foundation of a gated, UK-centric knowledge base and a strict, "provenance-first" architecture that always provides citations for its answers. This same principle—grounding every claim in a verifiable, authoritative source—is what will be required to make patient-facing AI truly safe.
A patient's Aide Health "Mirror" summary might list a new diagnosis or medication. The patient could then use a trusted, UK-aligned tool to get more context. For example, they could ask, "What is 'atrial fibrillation' according to a trusted UK source?" or "What are the key points about this new medication?" A provenance-first engine is the safest way to provide this post-consultation clarity, ensuring the explanation they receive is aligned with national guidance and free from the "hallucinations" of general-purpose AI.
FAQs
- Are these apps a replacement for my GP?
- No. Most symptom checkers, including K Health, explicitly state that they do not provide a diagnosis and that their advice should be followed by a consultation with a qualified clinician if symptoms persist.
- Which AI tools are actually linked to the NHS?
- The NHS 111 Wales AI assistant is live, and the NHS App "My Companion" is a central part of the 10-year digital plan. Other tools may be used in local NHS pilots.
- Which of these tools work internationally?
- Ada, Buoy, Healthily, MayaMD, and K Health are all designed for multi-country use, though their primary focus and verification may be in specific regions like the US or Europe.
Calls to action
- For patients: Choose tools that are transparent about their sources, provide clear safety-netting advice, and are linked to national health services where possible.
- For NHS leaders & insurers: Standardise the evaluation of these tools based on their accuracy, safety-netting performance, and data governance, and publish a clear list of "approved" AI tools for patients.