AI Scribe Vendor Risk: Plan for Supplier Restructures

AI scribe vendor risk became a news story on 5 November 2026, when the Australian Financial Review reported that Heidi Health, an Australian medical AI scribing company, plans to cut 18 per cent of its workforce. Two of its co-founders had announced their exit a day earlier. Whether or not your organisation uses Heidi, the story is a useful prompt. Any aged care or home care provider that relies on an AI documentation tool should know what happens to its notes, its support and its contract when a supplier changes shape.

Key Takeaways

A supplier restructure is not proof that a product will fail, but it is a clear trigger to review your contract, your data position and your backup plan.
Heidi Health told staff it will cut 18 per cent of its workforce, citing volatile sales and rising competition in the United States, according to the AFR.
Providers should be able to export their own records and keep notes readable without depending on the vendor's platform.
Leadership exits, layoffs, funding rounds and acquisitions belong on a vendor watch list inside your AI tool register.
Product changes after a restructure can alter what an AI scribe does, so route them through change control before staff rely on them.

What the Reporting Says About Heidi Health

The AFR's Emma Rapaport reported that Heidi co-founder and chief executive Thomas Kelly told staff in Melbourne the company would cut 18 per cent of its workforce. The announcement followed the departure of chief technology officer Yu Liu and chief financial officer Waleed Mussa, and it came two months after Heidi reached unicorn status through a $140 million funding round.

The report says Heidi grew from about 80 people at the end of 2024 to more than 600, and that its AI documentation tools are used in more than 2.8 million patient visits a week across 190 countries. The cuts are expected to hit its Melbourne research and development team, customer support and sales, and some roles will move offshore.

Kelly said he had been too optimistic about how quickly large hospital and health systems would adopt the technology. The article also notes that Epic, which supplies electronic health record systems, introduced its own scribe feature in February. Blackbird, an early investor in Heidi, told the AFR that the business is well funded and that it supports the decision. Canva had announced a restructure affecting 22 per cent of its workforce the day before.

Why a Supplier Restructure Is a Governance Matter

Nothing in the reporting suggests Heidi's product will stop working, and a company can restructure and remain a sound supplier. The governance point is narrower: a provider that adopts an AI scribe relies on another organisation's staff, product roadmap and finances to keep a clinical documentation tool running. When that organisation changes size or direction, the provider's risk position changes even if nothing looks different on screen.

The OAIC's guidance on commercially available AI products says privacy obligations apply to personal information put into an AI product and to output that contains personal information. It also says organisations should carry out due diligence before adopting a product. Due diligence is not a one-off event at signing. It needs repeating when the supplier's circumstances change.

What Providers May Notice When a Vendor Shrinks

These are signals to watch for, not predictions about any one supplier.

Slower responses from customer support, or a new account contact who does not know your setup.
Features that change, move to a higher plan or are retired without much notice.
New pricing, new contract terms or a different renewal process.
Changes to where data is processed or which subcontractors handle it.

The last signal connects to the practice described in our guide to managing AI vendor subprocessors, where a notification and review trigger matters more than a calendar date.

Check the AI Scribe Contract Before You Need It

Most providers sign these terms at onboarding and rarely read them again. When a supplier appears in the news, compare the contract with what your team needs. Look for clear wording on each of the following.

Notice of material change - how and when the vendor must tell you about changes to the service, its subcontractors or its ownership.
Data export - what formats you can export, how long an export takes and whether it costs extra.
Retention and deletion - how long audio, transcripts and draft notes are kept, and how deletion is confirmed.
Transition help - whether the vendor supports you for a set period if you leave or if the service winds down.
Data location and subcontractors - where information is stored and processed, and who else touches it.
Support commitments - response times and escalation contacts if something goes wrong.

The questions in our aged care AI procurement checklist work as a ready-made review list, and your own third-party and subcontractor management policy should say who owns the review.

Keep Notes Readable Without the Vendor

If AI-generated notes become part of a client's record, the provider remains responsible for keeping that record accessible, whichever tool produced the first draft. Confirm that finalised notes are saved in your own care management system and not only inside the scribe platform. Run a test export and open the file. Staff should also know which version counts as the finished entry in the progress notes record, and your retention approach should match our guide to retaining AI prompts and outputs.

Treat Product Changes as Change Control

Restructures often come with a shift in product focus, and that matters because the regulatory position of a scribe depends on what it does. The TGA's page on digital scribes says tools that only record, transcribe and translate conversations, without adding clinical content, are not medical devices. A scribe that suggests diagnoses, treatments or clinical conclusions is likely to be one.

A new feature can therefore change both the risk and the regulatory status of a tool you already use. Send feature announcements through the process described in aged care AI change control, and apply the sampling approach in AI output quality assurance before staff rely on the updated output.

Build a Vendor Watch Routine

Add a short vendor watch to your AI tool register. For each supplier, record the accountable owner, the contract end date and the data location, then add a line for organisational events. Useful triggers include:

Leadership departures or board changes.
Layoffs, restructures or offshoring of support teams.
New funding, acquisition or a change of ownership.
A major competitor launching a similar feature.
Changes to terms, pricing or subcontractors.

When a trigger fires, the named owner runs a short review. Read the source reporting, check the contract clauses above, confirm that an export still works, tell the governance lead and record the decision. Keep the response proportionate, because a pricing change needs less attention than a change in where data is processed. Your AI and automated decision-making policy is the right place to set that expectation in writing.

Plan for the Day the Tool Is Unavailable

A scribe can go offline for reasons unrelated to a restructure, such as an outage or a withdrawn feature. Staff need a written fallback covering manual notes, how to record anything missed and who approves a return to the tool. Link that fallback to your business continuity policy.

Loosely coupled tools make this easier, as the case for preventing vendor lock-in explains. If you decide to stop using a tool, the criteria in our guide to ending an AI pilot help you scale back without losing the record.

Related Resources

Common Questions About AI Scribe Vendor Risk

1. Does Heidi Health's restructure mean an AI scribe is unsafe to use?

No. The reporting describes a workforce reduction and leadership changes, not a product safety problem, and Blackbird told the AFR the business is well funded. The decision to keep using any scribe should rest on your own checks of contract terms, data handling and output quality.

2. Should aged care providers stop using AI scribes after news like this?

There is no automatic reason to stop. A supplier event is a prompt to run a short review and confirm your exit and fallback options. If the review finds gaps, fix them while the service is still running normally.

3. What should an AI scribe contract say about data export?

It should name the export formats, how long an export takes, any cost and how deletion is confirmed. Test the export once, because a clause on paper does not show whether the file is usable.

4. How often should we review AI vendors?

Review at onboarding, at each renewal and whenever a trigger event occurs, such as a restructure, an ownership change or new contract terms. Record each review in your AI tool register so the history survives staff turnover.

5. Are AI scribes regulated as medical devices in Australia?

It depends on what the tool does. The TGA says scribes that only record, transcribe and translate are not medical devices, while those that suggest diagnoses, treatments or clinical conclusions are likely to be. Check the vendor's stated purpose again whenever a feature changes.

Plan for Vendor Change Before the Headline Arrives

AI scribe vendor risk is manageable when the groundwork is done before a supplier makes the news. Confirm your contract terms, test your export, write the fallback and add a vendor watch line to your register. Start with the scribe supplier your staff use most, and record what you find.

AI Scribe Vendor Risk: What Heidi Health's Restructure Means for Care Providers