
Work Here?
Work Here?
Work Here?
Heidi Health builds a digital platform for managing medical documents and streamlining communication between general practitioners (GPs) and patients in Australia. It keeps referrals, prescriptions, test results, and follow-ups organized in one place and makes it easy to schedule appointments and message a GP. The product works through a web and mobile interface that users access via subscriptions or usage fees, giving teams and individuals quick, tap-away access to essential information. The company differentiates itself by focusing on the GP-patient workflow and convenient document management within the Australian market, delivered with a user-friendly experience and a culture of fast delivery and warmth. Heidi Health aims to reduce the need for physical waiting rooms, improve preventive and continuous care, and provide a clear, affordable way for practices and patients to manage medical information.
Industries
Consumer Software
Enterprise Software
Healthcare
Company Size
501-1,000
Company Stage
Series B
Total Funding
$91.3M
Headquarters
Melbourne, Australia
Founded
2019
See people who can refer or advise you
Help us improve and share your feedback! Did you find this helpful?
Total Funding
$91.3M
Above
Industry Average
Funded Over
4 Rounds
Industry standards
Hybrid Work Options
Paid Holidays
Paid Sick Leave
Professional Development Budget
Company Equity
AI scribe hallucinations in clinical notes: how to prevent them. July 16, 2026 A physician finishes a 14-patient day, opens the EHR, and reviews the AI medical scribe's output. The note looks clean. But buried in the assessment is a medication the patient never mentioned, and a diagnosis code that does not match what was discussed in the room. This is not a rare software glitch. It is a structural problem with how most AI scribes are built, and it has a name: hallucination. Healthcare AI hallucinations occur when a language model generates confident-sounding clinical content that has no basis in the actual conversation. The note looks complete. The language sounds clinical. The information is wrong. For a physician reviewing 20 notes at the end of a shift, that error is easy to miss and costly to ignore. What are AI scribe hallucinations in clinical notes? AI scribe hallucinations in clinical notes happen when the model fills gaps in the conversation with fabricated but plausible-sounding content. The trigger is usually ambiguity: a patient speaks quickly, a name is mispronounced, a symptom is mentioned in passing, or the acoustic environment introduces noise. Instead of flagging the uncertainty, the model infers, and sometimes incorrectly so. * Medication confabulation: The model generates a drug name similar to what was said but phonetically distinct (e.g., "hydroxyzine" instead of "hydralazine") * Diagnosis carries forward errors: Past medical history items bleed into the current assessment * Plan fabrication: The AI adds follow-up instructions or referrals not discussed during the visit * Demographic drift: Age, sex, or pronoun substitution when patient identity is not anchored in the transcript These are not edge cases. They represent the predictable failure modes of general-purpose language models deployed in high-stakes clinical environments without adequate medical tuning. Why AI scribe errors in clinical notes are a patient safety problem. The consequences of AI scribe errors in clinical notes extend beyond chart accuracy. A hallucinogenic medication can cascade into a prescription error. A fabricated referral can delay appropriate care. A misattributed diagnosis code can trigger a billing denial or payer audit. A 2023 study published in JAMA Internal Medicine found that large language models generated incorrect medical information in a meaningful percentage of clinical Q&A tasks, particularly around drug dosages, contraindications, and diagnostic criteria, exactly where clinical note hallucinations cause the most harm. A signed note is a legal document. If a hallucinated finding is signed and submitted to a payer, the physician has certified content that they did not generate. A physician seeing 20 patients per day who misses one hallucinated line per note accumulates documentation risk at scale, across a full patient panel, over months. Which AI scribes are most vulnerable to healthcare AI hallucinations? Not all AI scribes carry equal hallucination risk. The architecture matters. General-purpose large language models, trained on broad internet text and applied to clinical transcription, are most prone to healthcare AI hallucinations. They have no reliable mechanism for "I don't know." When audio is unclear, they fill the gap with statistically probable medical language. That language may be accurate. It may not be. Tools built on medically tuned models carry lower hallucination risk because their outputs are constrained to clinically grounded patterns rather than general language prediction. The distinction matters when evaluating AI medical scribe accuracy in real settings. A scribe who performs well in a clean studio environment may hallucinate frequently in a busy outpatient clinic, due to ambient noise, patient crosstalk, non-native speaker accents, and rapid dictation. Several Tier 1 competitors, Freed AI, Heidi Health, and Nabla, have published minimal clinical accuracy data. The market relies largely on user testimonials rather than independent validation. DeepScribe holds a KLAS score of 98.8/100, the highest ambient AI score recorded by KLAS Research, because it invested in third-party accuracy validation. That score carries weight with enterprise buyers for a reason. Notiro's medically-tuned AI scribe is built to handle multi-problem visit complexity and real exam room acoustic conditions, the two environments where general-purpose models are most likely to hallucinate. How to prevent AI scribe hallucinations: A practical framework. Preventing AI scribe hallucinations in clinical notes requires action at three levels: tool selection, workflow design, and physician review practice. 1. Choose a medically-tuned model over a general-purpose one. The first prevention step happens before the scribe enters an exam room. A tool trained on clinical language, one that understands the difference between a symptom, a finding, and a plan item, is structurally less likely to confabulate than a general-purpose model applied to medical transcription. Ask vendors directly: is the underlying model general-purpose or medically-tuned? Ask what clinical datasets it was trained on. Ask whether the tool has been independently evaluated for AI medical scribe accuracy. Vague answers should be treated as a warning sign. 2. Require uncertainty flags in the output. A well-designed AI medical scribe should distinguish between what it heard clearly, what it inferred, and what it could not capture. Notes that present all content with equal confidence provide no signal about where review attention is needed. Look for tools that surface low-confidence segments or leave explicit placeholders rather than silently filling gaps with fabricated text. 3. Audit high-hallucination sections first. The Assessment and Plan sections carry the highest risk of hallucinations; they require the model to synthesize and generate conclusions rather than transcribe. The Subjective section is the second-highest risk when the patient speaks quickly or uses non-standard terminology. Reviewing these sections first reduces cognitive load and focuses attention where errors are most likely to appear. Use this quick review checklist before signing an AI-generated clinical note: 4. Standardize the pre-visit context. One underappreciated hallucination driver is an information-sparse session start. When the ambient scribe begins recording without a patient history baseline, the model fills context gaps from training data rather than the actual patient record. Patient Intake AI, available in Notiro but absent from every other Tier 1 AI scribe competitor, addresses this directly. When the patient's presenting complaint, medication list, and history are captured before the visit starts, the scribe begins with a richer context. The note has less to infer and more to transcribe. Hallucination risk drops accordingly. 5. Build a post-visit review protocol. No AI scribe is error-free, and the review step is not optional. A structured protocol under 90 seconds, scan Assessment/Plan, check medication names, confirm diagnosis codes against what was discussed, catches the majority of hallucinated content before the chart is signed. The Mass General Brigham AI scribe study documented a 21.2% drop in physician burnout scores after 84 days of AI scribe use, conducted with a structured physician review workflow in place. The tool reduced documentation time. The protocol protected accuracy. How AI medical scribe accuracy affects billing, not just safety. Clinical note: hallucinations create a billing problem alongside the safety problem. A hallucinated finding in the Assessment section can generate an incorrect ICD-10 code suggestion, and an ICD-10 code that does not match the documented encounter is a claim waiting to be rejected. According to CMS, ICD-10 has over 70,000 codes, and CPT has over 10,000. Manual post-visit code selection under time pressure is already a systematic source of undercoding errors. An AI scribe that introduces inaccuracies into the note content it pulls from compounds. This problem. Notiro's ICD-10 and CPT coding pulls codes from the visit audio and the generated note. When the note is accurate, the code suggestions are accurate. When hallucinations corrupt the note, the downstream coding is corrupted too. Clinical note accuracy is not a documentation concern alone; it is a revenue cycle concern. How notiro helps prevent clinical note hallucinations before they reach the chart. Healthcare AI hallucinations in clinical notes are a predictable consequence of deploying general-purpose language models in environments that demand clinical precision. The physician who uses a tool that fills ambiguity with fabricated text is accumulating liability one signed chart at a time. Prevention requires the right tool architecture, a structured review protocol, and a pre-visit context baseline that reduces the model's inference burden. The hallucination problem does not disappear with better AI. It is managed with better AI and a disciplined clinical workflow. AI scribe hallucinations corrupt more than the note; they corrupt the codes, the billing, and the legal record the physician signs. Notiro's medically-tuned ambient scribe is built on clinical language, not general-purpose text prediction, and its Patient Intake AI gives the model a verified context baseline before recording starts. Start your free trial at notiro, no IT setup, no enterprise contract. Available now. Frequently asked questions. Find quick answers to the most common questions about Notiro - how it works, what it documents, and how it fits into your clinical workflow.
NHSE region signs AI scribe deal covering 15 trusts. Published 19 hours ago The NHS has launched its largest-ever regional deployment of ambient voice technology, covering 70,000 clinicians across 15 trusts and 1,239 GP practices. NHS England's Midlands team ran a competitive procurement process, selecting Australian vendor Heidi Health as sole supplier for the framework, which spans emergency departments, outpatient services, and primary care. The framework emerged from pilots at the Dudley Group Foundation Trust, which, according to Heidi, reduced emergency care documentation time by 80 per cent and cut a six-month rheumatology letter backlog to 14 days. Neighbouring providers expressed interest in replicating the business case and rollout, prompting NHSE's Midlands team to coordinate a single regional procurement route. Five trusts - Dudley Group, Sandwell and West Birmingham Hospitals Trust, the Royal Wolverhampton Trust, Walsall Healthcare Trust, and University Hospitals of North Midlands Trust - have begun deployment. Heidi declined to name the remaining 10 trusts expected to follow. The framework was opt-in, meaning it covers only those trusts that chose to participate and is closed to new joiners. 0 comments. There are no comments to display.
Conversion: the quiet gap where healthcare businesses bleed money. * July 9, 2026 Part six of a series on diagnosing growth instead of guessing at it. In the Healthcare Growth Equation (Growth = Demand x Trust x Conversion x Measurement), conversion is the variable where money leaks out speedily and invisibly. You've earned the demand. You've built the trust. Someone is interested, ready, even saying yes. And then, in the gap between decision and action, you lose them - and you almost never find out why. This article shows you how to bridge that gap. Splicemarketing'll explore what conversion really means, why friction is almost always the culprit, and how to create a smoother conversion process. Conversion isn't your win-loss rate. First, a definition, because the word gets used loosely. When I talk about conversion, I don't mean your sales conversion rate - deals won versus deals lost. I mean the moment between someone deciding they want what you offer and actually acting on it. For a clinic, it's the gap between "I should book that" and a confirmed appointment. For a healthtech product, it's the gap between sign-up and the moment usage becomes a daily habit rather than another abandoned tool. That gap is where most healthcare businesses bleed, and it almost always comes down to one thing: friction. If the path to action makes someone think harder, work longer, or take one extra step, you've lost them. You've experienced this yourself. Every time you've started an online form that felt more demanding than your tax return and just closed the tab, that was a business losing you due to friction in the conversion process. Friction is the enemy, and it hides in plain sight. The reason conversion problems are so expensive is that friction is invisible to the business creating it. You designed the booking form, so it doesn't feel like friction to you. In fact, gathering all that data may actually be helpful to your internal processes, making the extra boxes seem trivial. But to the person on the other side, every extra question, every system glitch, every moment of "Wait, how do I do this?" is a reason to give up - and they leave without telling you. The clearest example I know comes from clinical software. MedicalDirector runs systems used across the majority of Australian general practice. When AI scribes entered the market - genuinely useful tools saving GPs real time - the friction point was always the same. The GP would use the scribe during the consultation, then, once the patient left, they had to manually copy and paste the scribe's notes into MedicalDirector. One extra step sits between the technology and the workflow. So MedicalDirector built Smart Scribe, an integration connecting AI partners like Heidi Health directly into the patient record. The note goes where it needs to go, inside the system the GP already uses. No extra step. The market didn't fail because of that friction - it reorganised itself around removing it. The lesson scales down to the simplest clinic. Splicemarketing worked with a high-end practice that had full pricing published openly on its website. Splicemarketing understood the instinct to be upfront about pricing, but Splicemarketing also saw potential patients taking fright before they understood the value and leaving without booking. So Splicemarketing offered something of value to download that showcased the offerings and carried the pricing inside it, giving context for the cost. It was gated content, which meant patients gave their email address to download it, allowing the clinic to nurture them over time. Same information, completely different conversion outcome - because the friction was removed at the exact point people were dropping off. The two friction traps. Integration or process friction. This is when acting requires too much effort up front. In healthtech, that's months of implementation, IT sign-off, custom development; nobody has time to implement something even if it would ultimately save them time. In a clinic, it's the booking that requires a phone call during business hours, or a form that asks for too much information at the "getting to know you" stage. Workflow or expectation mismatch. When the thing works, but only if the person changes how they already behave to access it. Every step that asks someone to deviate from their existing habit is a step where adoption dies. The businesses that break through do the opposite of adding steps - they remove them, and they meet people where they already are. Heidi Health went directly to individual GPs and made it free to start: no IT approval, no implementation project. A clinician could try it in a single consultation, get value immediately, and tell a colleague. That was a conversion decision, not a marketing tactic. It reduced the effort required to say yes, turning one interested user at a time into momentum the business could scale. Conversion is hiding retention. Conversion isn't only about the first action. The same friction logic governs patient acquisition and retention alike. The nurture sequence that brings someone back next month, the follow-up after a first consultation, the recapture of a lead who went cold - these are all conversion, applied to the relationship rather than the first sale. In healthcare, lifetime value and repeat and referral behaviour often matter far more than first-visit conversion. A business obsessing over the first booking while ignoring what happens after it is only reading half the variable. The full picture runs from the first click all the way through to the patient who comes back, and refers a friend, because every step of that journey was made easy. How to tell if conversion is your weakest variable. Some honest signals that conversion is your binding constraint: If demand is bringing people to the door and trust is making them willing, but they still aren't acting, conversion is where your growth is leaking. What fixing conversion looks like. Conversion is the most mechanical of the four variables, which is good news - it responds quickly to deliberate work. Find the single change that removes the most friction for the least effort, and start there. In practice that means: * Shortening forms to only the essential fields * Testing the booking flow on mobile as ruthlessly as on desktop * Adding tap-to-call buttons and sticky enquiry buttons for mobile users * Regularly reviewing the patient journey to find and remove friction points * Building a real lead-handling and follow-up process, including SMS recapture for people who didn't book the first time. * Mapping the whole conversion journey end to end, from first interest through to retention, so no step is left to chance. It's worth the effort. Splicemarketing helped one company improve its conversion rate by 20% using a mix of these strategies. The bottom line. Conversion is where the work of demand and trust either pays off or disintegrates. The money you lose here is the hardest to see, because a patient who gives up doesn't fill in a form to tell you they've gone. Removing friction - at the booking, in the follow-up, across the whole journey from first click to returning patient - is often the fastest, highest-leverage growth available to a healthcare business. If you're generating interest that isn't converting into booked and returning patients, there's almost certainly friction you can't see from the inside. Book a free consultation with the Splice Marketing team.
Heidi launches a mic for AI scribe - why? By Adam Ang | March 26, 2026 | 4:25 AM Heidi has launched a wearable microphone designed to provide clinicians with reliable audio capture for AI-assisted documentation without using phones or laptops. WHAT IT DOES Intended to support the Heidi AI Scribe, the new Heidi Remote enables consistent audio capture in noisy environments. Weighing 21 grams and offering up to 14 hours of battery life, it can be clipped onto clinical attire for use throughout a full workday. The device operates independently of phone battery and Wi-Fi connectivity, allowing consultations to be captured offline and synced later when a connection is available. Heidi also said the device is built for frequent disinfecting and meets healthcare security standards, with all data encrypted. Heidi Remote will initially be available in Australia, New Zealand, the United Kingdom, the United States, and Canada, with the European Union to follow shortly. WHY IT MATTERS As the use of AI scribes is increasingly becoming a standard in clinical documentation, the "weakest link" has always been the device in the room, said CEO Dr Thomas Kelly. Heidi built its own microphone after finding that existing consumer and third-party devices place what co-founder and CTO Yu Liu described as an "integration tax" when applied to specialised clinical workflows. "We found it nearly impossible to achieve a seamless experience using third-party software stacks," he told Mobihealth News. Integrating external hardware would have required redesigning core software around vendor constraints while creating risks to stability as systems on either side upgrade, he added. Heidi said clinical environments were never designed for AI listening, pointing to ageing computer terminals, shared workstations, and personal smartphones that were not built to support AI-assisted care. The company opted for a wearable microphone instead of room-based or fixed systems to better support clinical work across varied and mobile settings. "Clinical care isn't static," Liu said, noting that room-based systems and telehealth setups can fail in edge cases, such as ward rounds, high-mobility clinics, or rural settings with limited connectivity. The CTO said the wearable form factor is intended to eliminate "workflow anxiety," ensuring consistent audio capture despite noise, movement, and inconsistent device performance. THE LARGER TREND Heidi Remote is the company's first venture into hardware as it evolves beyond AI scribe into a broader clinical workflow platform that includes documentation, evidence retrieval, and patient communications tools. It followed the launch of Heidi Evidence, introduced in February, an ad-free medical research tool on its AI Scribe, and Heidi Comms, which assists with coordinating patient communications across calls, bookings, reminders and follow-ups. ON THE RECORD "Clinicians shouldn't have to wonder if a room-based microphone will catch the audio or if a spotty internet connection will disrupt their notes... Whether a clinician is in a high-tech metro hospital or a remote clinic, Heidi Remote provides the same reliable performance, solving for mobility and connectivity gaps that fixed infrastructure simply cannot," co-founder and CTO Yu Liu told this publication.
Heidi Health, an Australian AI healthcare startup valued at $660 million, has launched the Heidi Remote, a 21-gram wearable device designed to record patient consultations. The device processes data on-device rather than in the cloud, ensuring patient information never leaves the clinic. Founded during the COVID-19 pandemic by Dr Thomas Kelly, Waleed Mussa and Yu Liu, Heidi Health has been used in over 118 million clinical sessions across 116 countries. The platform is now employed by half of UK NHS GPs, saving primary care doctors up to 90 minutes daily. The company recently raised $102.7 million led by Point72 Private Investments. Kelly emphasises that Heidi will never monetise user data, maintaining strict ethical standards whilst using AI to reduce administrative burdens for clinicians.
Find jobs on Simplify and start your career today
Industries
Consumer Software
Enterprise Software
Healthcare
Company Size
501-1,000
Company Stage
Series B
Total Funding
$91.3M
Headquarters
Melbourne, Australia
Founded
2019
Find jobs on Simplify and start your career today