InTouchLabs builds and proves patient and clinician focused AI, from voice and translation to computer use and evals, then puts it to work across UK general practice through InTouchNow AI.
patient calls handled every month, a live testbed no benchmark can imitate
GP practices already live with InTouchNow AI, the delivery business our research ships into
research programmes: voice, translation, computer use and evals
of the UK population is registered at a practice who uses our technology
InTouchNow AI has answered patient calls for NHS general practice since 2019, first with people, then with voice AI built once the operational job was understood. InTouchLabs is the next step: a first-of-its-kind research lab dedicated to NHS primary care, turning what 150+ practices teach us every day into the next generation of patient and clinician focused AI.
Research here is never speculative. Every programme starts with a problem observed in a real practice and ends as a capability in the delivery business, tested on live workloads, evaluated before it reaches a patient call, and supported after it ships.
Our questions come from reception desks, consulting rooms and call queues in UK general practice, not from a whiteboard.
Findings ship. The route from prototype to production runs through one company, so research reaches patients in months, not years.
Every capability is measured against our evals framework before deployment. Evidence first, then rollout.
Voice, translation, computer use and evals, studied through millions of real patient phone calls.
Most of the 200,000+ calls InTouchNow AI answers every month go exactly as designed. Research at InTouchLabs starts with the ones that do not: the accent a model mishears, the request no pathway anticipated, the conversation that switches language halfway through.
Every outlier is circled, replayed and turned into an evaluation. When a model can pass the calls that once broke it, the capability ships, and those outliers become the benchmark the next model has to beat.
Each programme pairs frontier models with the operational insight of a live call handling business, and each one feeds directly into what InTouchNow AI delivers to practices.
Speech models tuned for real patient calls: turn taking, interruptions, regional accents, noisy lines and callers who need more time. The aim is a conversation that works first time for every patient, whatever the hour.
Real time speech translation so patients can speak to their practice in the language they think in, starting with the languages UK practices ask for most, and staying accurate where accuracy matters.
AI that operates practice software the way a trained member of staff does: navigating clinical and booking systems, completing forms and leaving a full audit trail, so requests become actions without retyping.
Clinical grade evaluation of healthcare AI: benchmarks, adversarial testing and regression suites that measure safety, accuracy and fairness before any model handles a patient call. Nothing ships on a demo.
The lab's output is not a paper on a shelf. It is a patient getting through, being understood and getting what they called for, at any hour and in any language.
Patient demand does not stop at 18:30. Calls are answered around the clock, through the 08:00 rush, overnight, at weekends and on bank holidays, so an appointment route is always open.
Translation research means a patient can ask for help in the language they think in, with terminology that stays safe across languages.
Every request is reviewed by the practice team, and AI does not triage, diagnose, advise or prescribe. Research widens what is possible without moving clinical judgement.
200,000+ calls a month become structured, governed evidence. Findings are measured on live outcomes, so practices see what changed for their patients, not a demo.
AI does not triage, diagnose, advise or prescribe. Our research raises the standard of what it can safely do: answer, understand, translate, operate and prove itself, with the practice team in charge at every step.
One loop, continuously turning. What practices experience feeds what the lab studies, and what the lab proves ships back to practices.
Every programme starts with a real problem from practice: a missed call, a language barrier, a workflow that eats staff time.
We prototype with frontier models and the operational insight of a live call handling business serving 150+ practices.
Capabilities are measured against our evals framework and the standards NHS buyers expect, including the Digital Technology Assessment Criteria and the DCB0129 and DCB0160 clinical safety standards.
What passes ships through InTouchNow AI to practices, Primary Care Networks and partners across the UK, with support after go live.
The lab exists to serve UK healthcare, from national bodies and commissioners to the practices we already work with and the apps in patients' pockets. Wherever you sit in the system, there is a way in.
Evidence from live deployment to inform national thinking on AI in primary care: what is safe, what works and what patients actually experience.
Evidence on access, demand and AI safety that supports commissioning decisions at system scale.
Our customers and our testbed. Pilot new capabilities early, shape the research agenda and see findings before they become products.
Teams building apps that help people manage their health and navigate care: compare notes on voice, translation and safe AI behaviour.
Joint studies on voice AI, translation and evaluation methods, with access to real operational context.
Telephony, clinical system and infrastructure partners who want research backed AI in their own platforms.
Lab work is engineered to the same expectations as the delivery business: UK GDPR, clinical safety documentation to the DCB0129 and DCB0160 standards, and alignment with the Digital Technology Assessment Criteria. Patient data is handled under the practice's own lawful basis, and every research deployment is reviewed by the practice team it serves.
There is a lot going on in the lab right now. Pilot results, evaluation methods and new capabilities will be written up here as they land.
This is our first public release, and we want to hear from you, whether you commission NHS services, build healthcare technology anywhere in the world, or simply care about healthcare AI done properly.
GP practices, Primary Care Networks, Integrated Care Boards and NHS teams: discuss the research we do, join early pilots or explore what it could mean for your patients.
Start a conversationHealth systems, hospitals and teams building healthcare applications around the world: if our work on voice, translation, computer use or evals is relevant to yours, we would like to compare notes.
Reach out from anywhereResearchers, clinicians, engineers, students: if healthcare AI matters to you, engage with the lab. Challenge our thinking, propose an evaluation, or just follow the work as it is published.
Engage with the labIf you run a practice, commission services, research AI or build healthcare technology, we would like to hear from you. Use the form, and we will come back to you.
Prefer email? Write to us directly at hello@intouchlabs.ai.