Dr Kola Tytler (MBBS MBA MRCGP) | 15 July 2026 | 13 min read
The UK ambient voice technology market has moved past the demonstration phase. Every serious supplier can now show a note being generated from a recorded consultation. What increasingly separates them is regulation, distribution, integration, evidence and institutional credibility. Heidi, Tandem and TORTUS have each arrived at July 2026 by a different route, and the comparison is more useful than a simple ranking.
The market context behind the three strategies
It is worth setting out why this comparison matters now rather than a year ago. NHS England's April 2025 guidance on AI-enabled ambient scribing formalised expectations around clinical safety, data protection and human verification that had previously been handled inconsistently trust by trust. The NHS 10-Year Plan then set an explicit expectation that ICBs and providers adopt AVT at pace from April 2026. Sweden's Medical Products Agency's market surveillance initiative, which directly triggered Tandem's reclassification, has also made regulators elsewhere, including in the UK, visibly more attentive to how AI scribes are classified. All three suppliers are therefore competing inside a regulatory environment that has hardened considerably over the past twelve months, not the more permissive one they each entered.
Heidi: clinician adoption, then enterprise expansion
Heidi's defining strength has been recognition among individual clinicians. A standalone, low-friction product with highly configurable notes and documents let GPs and hospital doctors try it without organisational sign-off, and the Modality Partnership evaluation gave that adoption a documented evidence base in primary care: an initial 25-day pilot across 47 GPs and more than 2,800 consultations, expanding to more than 200 clinicians across 53 surgeries once results were reviewed. Heidi's own published figures describe support for over 1.5 million NHS appointments each month and use across a substantial proportion of UK GPs, alongside secondary care activity in settings such as same-day emergency care and outpatient clinics, where the company reports documentation time reductions in the region of 85 percent at some sites.
The reported Midlands NHS framework, positioning Heidi as sole supplier across 15 trusts, is the clearest sign yet of a shift towards enterprise-scale procurement, moving the company from a primarily clinician-adopted tool towards centrally commissioned regional infrastructure. Heidi remains registered as an MHRA Class I device for its current summarisation functionality and has stated publicly, through its own NHS compliance materials, that it is progressing towards Class IIa certification, though that certification had not been confirmed as achieved at the time of writing.
Tandem: a regulated platform distributed through Accurx
Tandem took a different route, treating documentation, coding and clinical decision support as parts of one regulated workflow from early on rather than adding regulation retrospectively. Its AI Medical Scribe achieved EU MDR Class IIa certification on 21 May 2026, following independent notified body assessment, after Sweden's Medical Products Agency's market surveillance initiative found the product's prior Class I self-declaration incompatible with its intended purpose and risk profile under EU MDR Rule 11. Notably, the resulting nonconformity was categorised as minor rather than major, on the basis that Tandem had already been actively preparing for Class IIa certification at the time of inspection, which suggests the reclassification was already underway rather than purely reactive.
Its Coding Assistant reached the same Class IIa milestone earlier, in February 2026, and the company has since added a Class IIa clinical decision support feature to the same regulated platform, giving it three certified components covering the encounter from note generation through to coding and point-of-care guideline queries. Tandem's relationship with Accurx gives it a route into infrastructure already embedded across NHS primary care, and the company emphasises direct integration into patient records across more than 100 European EPR systems. It is used by more than 5,000 care organisations across Europe and has expanded through partnerships including Humanitas in Italy, alongside a reported four-year multi-trust NHS agreement giving around 10,000 clinicians access to its technology in the UK specifically.
TORTUS: NHS evidence and UK regulatory positioning
TORTUS became the first ambient voice technology in the UK to achieve UKCA Class IIa certification, assessed by UK Approved Body Scarlet (8536) on 24 June 2026, having previously held MHRA Class I registration since a comparatively early entry into the category. The company has built its NHS presence around large-scale clinical evaluation, including a multi-site study spanning nine clinical locations and reportedly more than 17,000 patients, independently evaluated, which it credits with informing the government's own NHS Ten Year Plan thinking on AVT. TORTUS positions its Class IIa status as recognition that ambient scribing composes clinical documentation rather than merely transcribing speech, which it argues places it closer to Class IIa functionality by intended purpose rather than marketing convenience. The company also reports integration across the majority of NHS EPR systems and telephony services including X-on and Avaya, and describes over 2.5 million consultations recorded across the NHS to date.
Comparing the three
| Dimension | Heidi | Tandem | TORTUS |
|---|---|---|---|
| Principal route to market | Clinician-led adoption, now expanding to NHS frameworks | Accurx distribution and direct NHS multi-trust agreements | NHS institutional deployment and formal evaluation |
| Primary care presence | Strong (Modality Partnership and wider GP adoption) | Growing via Accurx integration | Present via partners such as X-on |
| Current regulatory status (reported) | MHRA Class I; Class IIa reportedly in progress | EU MDR Class IIa (scribe, coding assistant, decision support) | UKCA Class IIa (scribe/decision-support functionality) |
| Certifying regime | UK MHRA registration | EU MDR, notified body assessed | UK UKCA, Approved Body 8536 |
| EPR integration claims | Configurable, expanding | Over 100 EPR systems across Europe | Integrated across majority of NHS EPR systems (company claim) |
| Coding functionality | Not a headline feature | Dedicated Class IIa Coding Assistant | Codes suggested alongside notes |
| Published NHS evidence | Modality Partnership evaluation; wider Impact Report data | Capio and other European evaluations | Nine-site, 17,000+ patient evaluation |
| Recent institutional deal | Reported Midlands 15-trust framework | Reported four-year, ~10,000-clinician NHS agreement | Multiple NHS trust deployments |
| International footprint | Global, strong UK and Australian presence | 5,000+ organisations across 10+ European markets | Primarily UK-focused |
What buyers should read into scale claims specifically
Each company reports its scale slightly differently, and it is worth being precise about what each figure actually measures before comparing them. "Consultations supported per month" is not the same as "active weekly clinicians," which is not the same as "licensed clinicians," which is not the same as "trusts under contract." Heidi's monthly appointment figures are large, but they describe volume of use among an existing, engaged clinician base rather than breadth of formal institutional deployment. Tandem's organisation count reflects European scale more than UK-specific depth. TORTUS's consultation total is smaller in absolute terms but sits alongside the most extensively published independent evaluation of the three. None of these figures is more or less honest than the others, but none of them answers the same question, and procurement teams comparing suppliers side by side should ask each company to report the same metric before drawing conclusions from a table like the one above.
Where Heidi may currently hold an advantage
Clinician familiarity built over roughly two years of grassroots use is not easy to replicate quickly, and it gives Heidi a base of internal advocates inside organisations before any formal contract is signed, which materially eases local adoption once a framework activates. Cross-market scale, with claimed usage figures well beyond the UK, and a demonstrated ability to convert individual adoption into organisational demand, as the Modality and now Midlands examples suggest, are genuine strengths that neither competitor has yet matched at the same scale within UK general practice specifically.
Where Tandem and TORTUS have created pressure
Both companies now hold independently assessed Class IIa certification, which Heidi does not yet hold for its current functionality. Both have built clearer regulated-platform narratives aimed directly at NHS procurement and clinical safety teams, and both emphasise formal evidence generation and direct clinical-system integration as core to their pitch, rather than as an add-on to a clinician-adoption story. Tandem's willingness to publish the detail of its own regulatory reclassification, including the Swedish inspection findings that prompted it, is also a distinctive piece of transparency that sets a bar other suppliers, including Heidi, may increasingly be expected to match.
Why this should not be read adversarially
NHS organisations have genuinely different requirements by setting, specialty and existing IT estate. A community mental health trust and a large teaching hospital emergency department are not shopping for the same thing, even if both are technically buying "an AI scribe." A higher device class does not automatically mean better note quality or higher clinician satisfaction, and a larger user base does not automatically mean a safer product. Competitive pressure between these three, and others entering the market, is more likely to raise standards across the category than to produce a single winner that suits every NHS use case. The right question for a given trust or ICB is rarely "which of these three is best" and much more often "which of these three fits our specialty mix, our EPR estate and our governance capacity best, right now."
Conclusion
Heidi, Tandem and TORTUS are not simply competing to generate the best consultation note. They are competing to define the infrastructure, regulatory expectations and clinical workflows that will surround ambient AI across the NHS for the next several years. Heidi's Midlands framework, if it proceeds as reported, is best read as its answer to a question Tandem and TORTUS have already been asking through certification and institutional deployment: how do you turn clinician popularity into governed, scalable NHS infrastructure, without losing the grassroots trust that got you there in the first place.