A patient walks in and describes something complex and human. Pain they live with, anxiety they carry, or a child's symptom a parent can't quite put into words. The clinician listens, asks the right questions, and builds a picture with careful follow-up. They want to be fully present. But part of their attention is already on the screen, because every minute in the room also has to become a note, an updated health record, a referral letter, a billing code.

That work does not disappear when the patient leaves. It has to go somewhere. Either it intrudes on the consultation and pulls the clinician's eyes from the person in front of them, or it follows them home.

Clinicians are estimated to spend more than two hours a day on documentation, much of it after the last patient has left. The profession even has a name for it: pyjama time, the paperwork finished in bed. Burnout is rising, and most clinicians point to administrative load as a primary driver. Health systems lose good clinicians to early retirement, career changes and sheer exhaustion.

For decades the only help was dictation software that turned speech into text but still left clinicians to organise, structure and edit every word. The burden barely moved.

And documentation is only the most visible part of the load. The same clinician still has decisions to weigh, evidence to chase down, and follow-ups to coordinate, all of it competing for the attention the patient came for.

Enter Heidi.

Heidi is best known as an ambient AI scribe: it listens during a consultation and turns it into an accurate, structured clinical note, lifting the documentation that consumes so much of a clinician's day. That is how Heidi started, and it is still how most clinicians first meet it. But the scribe is the entry point, not the destination. What began with the note is becoming a care partner for the whole clinical day, generating referral letters and task lists from a single session, recalling a patient's history across encounters, surfacing answers at the point of care, and letting clinicians work with all of it through a simple command layer.

What makes Heidi distinctive is who it serves and how it spreads. Where most tools have focused narrowly on physicians in large hospital systems, Heidi works for every kind of clinician: general practitioners, psychiatrists, nurses, allied health professionals, dietitians, even veterinarians. Its template library, built and refined by clinicians themselves, adapts to specialities, workflows and geographies in a way that top-down enterprise tools cannot.

Clinicians have responded at scale. Heidi reports that it now supports more than two million consultations every week, across well over a hundred countries, and a 2025 KLAS Research spotlight found every customer surveyed would buy Heidi again.

Heidi was founded in 2019 by Dr Thomas Kelly, Waleed Mussa and Yu Liu. Tom is a former surgical resident who has spent close to a decade building healthcare software, and a large share of the team come from clinical backgrounds, which shows in how naturally the product fits the way clinicians actually work.

In October 2025 the company raised a US$65M Series B led by Point72, with Blackbird and existing investors including LocalGlobe, Headline and Archangel.

Enterprise sales in healthcare is notoriously slow: months of procurement, bake-offs across dozens of physicians, deep integration testing. Heidi took the opposite path, and not by choice. The free, frictionless model was forged in the near-death moment described below. Having learned that top-down sales of a product nobody used almost ended the company, the team made the scribe free, stripped out every barrier to starting, no onboarding, no IT department, no approval, and treated the compute cost as its marketing budget. As Tom frames it, the real question is not who signs the contract, it's who loves your product.

It worked because clinicians did love it. A single doctor tries Heidi in their own clinic, saves an hour or two a day, and tells a colleague. That colleague tells their department, and the department tells the health system. By the time an enterprise conversation begins, Heidi already has passionate internal champions and real usage data behind it. This is not how most enterprise software spreads: clinicians routinely bring Heidi into their workplaces themselves, championing it from the inside and pressing their institutions to make it official.

The motion is organic. Heidi's intuitive interface, instant onboarding and distinctive brand identity create genuine word-of-mouth within clinical networks, and the result is a distribution engine of unusual efficiency. Heidi has spent far less than its better-funded competitors on paid acquisition while growing at a remarkable pace. The people who use the product have become the people who sell it.

Ambient scribing itself is commoditising, so the real question is what compounds. For Heidi, it is the clinician community, and how deeply the product serves it. Clinicians build, share and refine documentation templates across specialities, and the most influential become, in effect, product designers whose work lifts the value of the platform for everyone.

This is a flywheel. More clinicians bring more templates and more usage, which makes the product better and lets Heidi iterate faster than competitors relying on top-down design, which in turn draws more clinicians. The same loop feeds accuracy: large-scale, real-world use across primary care, psychiatry, allied health, aged care and veterinary medicine, segments incumbents have largely ignored, gives Heidi a data advantage that is hard to buy and harder to copy. It is a moat built by users, not bought with capital.

The community is one kind of moat. How deeply Heidi embeds in a clinician's day is another, not because breadth is a moat in itself (it isn't, competitors can build features too), but because the more of the day that runs through Heidi, the harder it is to remove. The scribe was the wedge. Heidi is now building toward an AI care partner for the whole clinical day. Evidence brings trusted clinical answers to the point of care, drawing on sources like the BMJ and MIMS, so a clinician can look something up without leaving the consultation. Comms handles patient communication, fielding calls, reminders and follow-ups so practices stay responsive without piling more on the front desk. And Remote, a 21-gram wearable microphone that works offline, takes Heidi off the desktop and into wards, theatres, rural clinics and emergency departments, wherever care actually happens.

As these connect, they compound. A clinician who documents, looks things up, communicates and captures care all in one place has little reason to leave, and is worth far more to Heidi than one who only takes notes. The category is chasing the same roadmap, coding, agents, follow-up, so the real differentiator is not who ships the most features but who can get clinicians to actually use them. That is the hard part in healthcare, and it is the one thing Heidi has already proven it can do.

The US ambient-scribe market is fiercely contested, with deeply integrated, heavily funded incumbents like Abridge, Ambience, Nuance DAX and Suki wired into the major electronic health records. Heidi's insight was that most of the world's clinical care happens outside those tightly integrated systems, in small practices, allied health, aged care and public health services across the UK, Europe, Canada, Australia and beyond, the places integration-heavy vendors are not built to reach. Its frictionless adoption and clinician-built templates make it well suited to exactly these fragmented markets, and it already has real momentum across the NHS and Australasian health systems. From there it is expanding on three fronts: North America, building local teams across the US and Canada; Europe, where the acquisition of UK clinical-AI company AutoMedica is accelerating the push; and Asia-Pacific, where a new Singapore base anchors its move into Southeast Asia.

What is most striking is that the same model is now winning inside the US too. In April 2026, Beth Israel Lahey Health, one of the largest health systems in Massachusetts, chose Heidi to roll out ambient AI scribing to all of its physicians. Heidi does not need to win every US bake-off, but proof that clinician-led adoption can carry it into a marquee American system, on the incumbents' home turf, materially widens the prize.

Healthcare is one of the hardest sectors to build for, and clinicians will not adopt tools built by people who do not understand their world from the inside. That is exactly where Heidi's founders sit. Tom Kelly trained as a vascular surgical registrar before leaving medicine in 2021 to build Heidi, and felt the problem himself. In a system stretched beyond capacity, a doctor can still only see one patient at a time.

What gives us most confidence is what the team did when Heidi nearly failed. The scribe was not the first product. Heidi began in 2019 as Oscer, an AI tutor for medical students, and during the pandemic became a sprawling AI telehealth platform sold top-down to hospital executives. Contracts were signed and projections rose, but clinicians would not use it. It basically became “shelfware” and by late 2023, after layoffs and with months of runway left, the company was “entirely cooked” according to Tom. Rather than push harder on a product nobody loved, they stripped it back to one principle, do not confuse the buyer with the user, killed the platform and rebuilt around the single thing doctors would reach for unprompted: writing the note. Walking away from signed revenue because the people who mattered would not use the product is rarer and harder than the pivot itself, and the clearest sign we have seen of a team that listens to its users above all.

The category is crowded. Dozens of well-funded vendors worldwide are building variations of ambient documentation and clinical-workflow AI, competition is fierce in the US, and differentiation is narrowing as almost everyone chases the same roadmap: coding, pre-charting, clinical agents, task execution. If the features are converging, the fair question is what protects Heidi.

Our answer is that the advantages that matter sit beneath the features, not in them. Heidi's distribution is hard to copy: a free, clinician-loved product that spreads by word of mouth rather than procurement. Its community and usage data compound into a product that adapts to specialities and geographies faster than top-down rivals can. And as Heidi extends across the clinical day, it becomes harder to displace, not because it has more features, but because more of a clinician's work runs through it. In a converging market, the winner is not whoever ships the most, but whoever clinicians actually choose, and Heidi has shown it can be that company even in the US, on the incumbents' home ground.

What drew us to Heidi was a rare combination: a product clinicians genuinely love; a distribution engine that turns that love into scalable growth; a community and an expanding set of products that embed it deeper into the clinical day and make it harder to displace; and a team with the clinical insight and judgment to keep building what clinicians actually use.

The timing matters too. The ambient scribe is at an inflection point, adoption is accelerating, and the category is shifting from note-taking towards the interface through which clinicians manage their whole working day. Heidi is moving at a pace that is rare in any market and rarer still in one as regulated and high-trust as healthcare, shipping and iterating fast enough to lead that shift rather than chase it.

We are backing Tom, Waleed, Yu and the Heidi team because they have earned the trust of clinicians in more than a hundred countries, because they have the judgment to build only what clinicians will adopt, and because they are building something that makes the lives of healthcare professionals, and the patients they care for, better.