Dr Thomas Kelly – who traded life as a vascular surgeon to co-found Heidi in 2019 – told the Herald the Catalyst funding was not private credit because it was specifically provided for acquiring customers. There were no debt covenants attached.
If it cost US$100 to acquire a customer, Heidi would then pay that amount back plus a set margin – for a total of $110 or $120 – with Kelly’s firm keeping 100% of the cashflow from that customer.
It was critical to have a non-dilutive source of funding to protect shares owned by employees – which were an incentive for staff to stay at a time when the Albanese Government’s capital gains tax overhaul had seen some Heidi staff asking about moving overseas to escape what they saw as a “tax grab”.
Where the money’s going
Heidi will use its new venture capital to expand its global presence and its suite of products, which now includes tools for integrating patient records and its Evidence medical explainer platform, which lets a GP – or anyone ask a question.
“Ask mid-visit. Heidi answers, scores and drafts what comes next,” the company says.
Kelly says Evidence has two key differences with general AI chatbots: it only taps trusted sources, with citations provided for every line, and it carries no advertisements. Evidence has fielded 10 million queries since it was launched in March.
Some of the new funding will go towards a new translator product, still in the early stages of development.
Kelly said it would have the same features as his firm’s AI scribe, but add translation features if a patient spoke, say, Chinese or Samoan and had no English.
“When I used to work in hospitals, I used Google Translate more than once,” Kelly said. The aim was to develop a translator that could handle both clinical grade medical notes and accurate language translation.
“If you’re in an ED [the emergency department] at 3am, whether someone describes a pain as ‘sharp’ or ‘dull’ can be the difference between whether a surgeon decides to operate.”
Growth surge
Heidi says its AI Care Partner (its core tool for capturing a doctor-patient chat, then summarising it as medical notes and drafting any required letters) has now been used for 175 million patient visits, up from 73 million at the time of its October 2025 Series B round.
Annual recurring revenue (or ARR – the monthly subscription revenue extrapolated over the next 12 months) has grown from US$1m to US$50m (as of April 2026) in just two years, the firm says.
Its contracts include a sole-supply deal with the UK’s National Health Service (NHS) for NHS Midlands, which includes licences for 70,000 clinicians serving about 11 million patients.
Heidi says it is the largest AI clinical documentation procurement in NHS history. It followed pilots that cut emergency documentation time by 80% and a six-month rheumatology backlog to 14 days.
Australian public health system rollouts including the Royal Children’s Hospital Melbourne, the Children’s Health Queensland Hospital and Health Service and Metro South Health (the largest regional division of Queensland’s public health system). Heidi has also expanded into the US, Canada, the Middle East and Asia.
Founder sees over-regulation threat
In Australia, Prime Minister Anthony Albanese recently created the Office of AI.
“It’s sensible for the Government to have some dedicated oversight,” Kelly said.
“But I’m worried about the risk of over-regulation of AI.
“If you make something perfectly safe, you stifle innovation.”
It was early days. In the past, regulation had been informed by actual experience.
“We had cars and people had accidents before we had seatbelts,” Kelly said.
Will young doctors lose a vehicle for critical thinking?
University of Otago associate professor Dr Ben Gray told the Herald that while he had never used Heidi’s scribe personally, he was worried that AI tools would result in info dumps in a context where notes had to be succinct and to the point.
He also worried that a generation of doctors would come through with no note-taking skills.
“It’s important that clinicians learn to synthesise their thinking,” Kelly responded.
His AI product also allowed clinicians to add their own notes – or not use AI at all if that was their choice or their patient’s choice.
A Hawke’s Bay trial had shown that Heidi’s scribe could cut the time it took to process a patient’s documentation from 17 to four minutes (Gray said the Hawke’s Bay pilot, which involved 10 clinicians over two months, should have been larger and longer).
Kelly said a doctor should always check the AI notes, as they should if they wrote them themselves.
“Doctors can hallucinate [make mistakes] too,” he said.
Kelly said AI note-takers were also proving more neutral. “If you think there’s a certain condition, you’re likely to over-emphasise it,” he said.
AI backlash?
He was not concerned about the growing public backlash against artificial intelligence.
“So far we haven’t seen too much of it,” Kelly said.
“A lot of people have experienced a clinician using an AI scribe and had a better experience,” he said.
It allowed a doctor to focus on their patient rather than their keyboard.
Regardless, the power should be with the patient, Kelly said.
“I still believe in patient choice. There should be a consent process. The patient should always choose.”
Similarly, he said he had no issue with clinicians who chose not to use AI.
“There will always be some doctors who never use [AI] scribes, or even electronic records,” he said.
“My GP used cue cards into the 2000s.”
Chris Keall is an Auckland-based member of the Herald’s business team. He joined the Herald in 2018 and is the technology editor and a senior business writer.