Heidi

Heidi started as a bet by a doctor who watched patients get hurt by missed and delayed diagnoses, and it has grown into one of the fastest-moving names in clinical AI. It listens to a consultation, writes the note, applies the codes, and handles the follow-up, so clinicians can look at the patient instead of the keyboard.

Founded in Melbourne in 2019 by vascular surgery resident Dr Thomas Kelly with Waleed Mussa and Yu Liu (the product first shipped as Oscer), the company now powers more than two million patient visits a week across 116 countries and 110-plus languages. In October 2025 it raised a $65M USD Series B led by Point72 and crossed a $465M USD valuation, and it has since started buying and building its way past the scribe into evidence and patient comms.

Heidi is an ambient AI platform for clinicians. A doctor speaks naturally during a visit and Heidi produces a structured clinical note, the billing codes, referral letters, and the tasks that follow, with no cleanup after. It calls itself an AI Care Partner rather than a scribe on purpose: the pitch is that it covers the whole arc of a visit, not just the transcript.

The company says clinicians save roughly six minutes per encounter and about ten hours a week, and that it has returned more than 18 million hours to frontline staff in 18 months. Those are Heidi's own figures, but the volume is real: over two million consults run through it weekly.

The problem Heidi attacks is administrative drag. Documentation has become the tax on every visit, the largest single driver of clinician burnout, and the reason good doctors leave the bedside for the desk. Heidi's framing is blunt: every minute on notes is a minute not spent with a patient.

So it targets the whole task list around a visit, not just the note. Care-gap identification, coding and clinical documentation improvement, revenue-leakage capture, referral triage, repeat prescriptions, and chronic-care remote monitoring outreach are all things it pitches to health systems as recoverable time.

Heidi is built to span the whole market, from a single family doctor to an NHS trust. The same product that a solo clinician downloads for free is the one health systems roll out across 200-plus specialties, from emergency and psychiatry to oncology and dentistry.

That reach is the selling point for enterprise buyers: one AI documentation tool across multidisciplinary teams instead of a patchwork. It has also pushed into elite sports medicine, pitching high-performance units on the same fast, citation-backed workflow.

Heidi's platform is really three tools that share one workflow. Scribe is the core: live transcription that turns a conversation into a note, codes, and tasks. Evidence puts citation-backed clinical answers on tap at the point of care, built out with the 2026 acquisition of UK startup AutoMedica. Comms handles patient scheduling and voice-to-voice communication.

There are two smaller pieces worth knowing. Dictate is a free desktop tool that drops your voice into any app, from the EHR to email, in over 90 languages. And Heidi Remote, launched in March 2026, is a 21-gram clip-on wearable that captures the consult without a phone on the desk.

Heidi runs a freemium model, and it means it: the Free tier gives every clinician unlimited AI documentation and unlimited clinical evidence with citations, at no cost, forever. That free base is how it found product-market fit and how it lands inside a practice before selling up.

Paid plans are per user per month, billed annually, and climb from the individual Clinician plan into Practice for teams and a custom Enterprise tier that adds SSO, EHR integration, centralized billing, and dedicated success. The catch that reviewers flag is price: at $180/user/month for Practice, Heidi is no longer the cheap option among AI scribes.

AI medical scribes are a crowded category, with Abridge, Nabla, Suki, DeepScribe, and Microsoft's Nuance DAX all chasing the same clinicians. Heidi's answer is to stop being just a scribe: the AutoMedica acquisition and the Evidence and Comms launches are a deliberate move from transcription into clinical reasoning and patient communication.

Its other edge is distribution. A genuinely free tier plus deep NHS penetration (the company says over 60% of NHS GPs use it) gives Heidi a bottom-up wedge that a pure enterprise sale can't easily match.

Heidi leans hard on measurable outcomes, and the specifics are its strongest sell. It claims 60 to 80% adoption in the first month against 20 to 40% for competitors, 61% less documentation time outside sessions, and 82% faster referral preparation.

On the money side it points to real deployments: one urology practice recovered around $121,000 in productive clinical time in 12 to 16 weeks, with reported ROI multiples of 5x to 10x on setup costs. These are the company's own case-study figures, so read them as its best cases, but they are grounded in named customers, not vague claims.

The praise is consistent: clinicians single out the multilingual documentation across 110-plus languages, the flexible template system, and a clean interface that takes almost no learning curve. One emergency physician described cutting 90% of documentation time and only tweaking about 5% of the generated note.

The criticism is just as consistent, and it is about reliability. Third-party reviews in 2026 report lost recordings and dropped sessions, with some clinicians saying they lose several a week, plus trouble keeping multiple problems separate in long, complex visits. Price comes up too. It is a fast product growing faster than its stability, at least in the eyes of some heavy users.

Heidi closed a $65M USD Series B (around $100M AUD) in October 2025, led by Point72 Private Investments, at a $465M USD valuation. That brought total funding to nearly $100M USD and turned a Melbourne scaleup into one of Australia's more valuable health-AI companies.

The cap table is deep for its stage: Blackbird led the original 2021 seed when the company was Oscer, and Headline, Latitude (Phoenix Court), LocalGlobe, HESTA, Possible Ventures, and Archangel have all backed it since. The money is going toward the UK, US, and Canada, including doubling UK headcount to keep up with NHS demand.

Heidi has the rare combination of a free product clinicians actually adopt and an enterprise story that health systems will pay for, and it is spending its Series B to press that advantage across the UK, US, and Canada. The move into evidence, comms, and hardware is a real attempt to become the clinical operating layer, not just the note-taker.

The risk is execution at that speed. The same growth that drove two million weekly consults is also behind the reliability complaints, and in healthcare a dropped session is not a minor bug. The next chapter is whether Heidi can keep shipping fast without breaking the trust the whole pitch depends on.

Heidi was founded by clinicians and still recruits them onto the product and engineering teams, to the point that some AI engineering roles ask for a medical degree. The team mixes doctors, engineers, designers, scientists, and mathematicians around one line: double the world's healthcare capacity without losing the human part of care.

The values read like a healthcare startup that knows the stakes. Live Forever (every release is measured and safe), Small Cuts Heal Faster (ship fast, don't break what people depend on), Practice Ownership, and Make Others Better. Work is hybrid at three days in the office, with equity from day one and up to four weeks a year working from anywhere.

Heidi discloses salary ranges on a chunk of its roles, mostly for its US and Canadian teams and denominated in USD and CAD. Engineering sits at the top of the disclosed bands, with sales and operations spread wider depending on level and location.

Equity is the constant. Roles across functions advertise stock from day one, and the company frames it plainly: when Heidi wins, you win. Sales roles carry commission on top of base.

Heidi handles clinical audio and health records, so trust is table stakes, and it stacks certifications accordingly: HIPAA, GDPR, SOC 2 Type II, ISO 27001, ISO 42001, Cyber Essentials Plus, NHS requirements, and the Australian Privacy Principles. It says consultation audio is deleted right after the note is generated, never stored permanently, and that it takes a zero-tolerance stance on third-party access.

The legal entity behind it is Heidi Health Trading Pty Limited, an Australian company (ABN 84 649 783 871). Data residency follows the market: EU and EEA data stays in region under GDPR, and Australian patient data stays in Australia.

The last year has been busy. October 2025 brought the $65M USD Series B and the $465M USD valuation. February 2026 brought a strategic pivot: Heidi acquired UK clinical-AI startup AutoMedica and launched both Evidence and Comms, moving from scribe to full care partner.

Then in March 2026 it shipped Heidi Remote, a 21-gram wearable recorder, its first hardware. The through-line is a company trying to own more of the clinical workflow than any single scribe competitor.