AI in healthcare teams

Rolling out a new tool in a small healthcare team

Treat a new system or AI tool as a small project with a named owner, not a switch you flip on a Monday. Decide who makes the call, try the tool with two or three people first, train everyone before it goes live, check in after two weeks, and agree in advance how you would go back. For a general practice adopting a digital health technology, RACGP criterion F10.A says the practice "has a documented process for assessing, costing, implementing and managing digital health technologies", so write the plan down. The same steps work for an NDIS service, with extra time for support workers who are less confident with technology.

Who should decide whether we adopt a new tool?

One named person, after hearing from the people who will use it. In a small team that is usually the practice manager, the owner or the operations manager. Their job is to make the decision, own the plan and be the person staff go to when something is not working.

If the tool is an AI tool, the decision also belongs in your AI policy. For a general practice, AGPAL's guidance on AI in general practice says AI's use "must be approached with care, vigilance and strong clinical oversight". In its summary of the Australian Commission on Safety and Quality in Health Care's principles for safe AI use, AGPAL lists reviewing the evidence for a tool's effectiveness and reliability, and being aware of its limitations, before use.

How can we encourage staff to accept a new tool?

Ask the team before you decide, not after, and keep asking once the tool is in. The RACGP's F10 – Digital health technologies suggests a practice could seek input from the practice team in the initial stages of implementing a new digital health technology, explain the benefits and expected outcomes to the team, address training and support needs, and encourage ongoing feedback. The pilot, the training and the two-week check-in below are the places to do it.

What does a rollout plan look like?

A page or two that anyone in the team can read. The example below shows one way to set it out for a small team. It is fictional, and it describes a plan, not a result.

A six-person general practice (two GPs, a practice nurse, two receptionists and a practice manager) plans to replace its paper recall list with a recall feature in a new online booking system.

Rollout plan: new recall feature

  1. Who decides. The practice manager owns the decision and the plan, and before signing asks each person what they need the recall list to do and what worries them about the change.
  2. Before go-live. The plan records what the change affects: who sends recalls today, which steps change, what the new system costs to run, who supports it, and what happens to recalls already on the paper list.
  3. Pilot. One receptionist and one GP use the feature for two weeks. Everyone else keeps the paper list, so nothing is missed if the pilot stops.
  4. Training. Before the pilot, the vendor runs a short session for the two pilot users. Before the full rollout, the pilot receptionist shows the rest of the team, using the practice's own recall types. A one-page guide sits at each desk.
  5. Check-in after two weeks. The pilot users and the practice manager meet for twenty minutes. They look at whether recalls went out, what took longer than expected, and what questions came up. They decide to go ahead, change something first, or stop.
  6. How to roll back. The paper list is kept up to date until the full team has used the new feature for a month. If the practice stops, it goes back to the paper list and the practice manager asks the vendor how to export the recalls entered so far.
  7. After go-live. The practice manager puts a review of the feature on the agenda for the team meeting three months later.
Synthetic example. A fictional practice and a fictional plan, not drawn from any real practice or client, and not a record of any result. An NDIS service could use the same outline for a new rostering or progress notes app.

The outline draws on prompts F10 sets out under "Assessing, costing, implementing and managing digital health technologies". Among them, a practice could consider the impacts on practice workflows, role responsibilities and readiness of the practice team, identify resource requirements, consider direct and indirect costs, and "define how the technology will be implemented, reviewed and managed over time, including how risks and issues will be identified and addressed".

What does RACGP F10 expect?

A documented process, and an assessment before any new technology goes in. F10 is part of the RACGP Standards for general practices (6th edition). Criterion F10.A says the practice "has a documented process for assessing, costing, implementing and managing digital health technologies, including consideration of their potential impacts on the practice, members of the practice team and patients". Under "Meeting these criteria", it says any proposal for new technology includes "a systematic, operational assessment of its impact before its implementation", which "could include workflow changes, readiness of the team, resource requirements, costings and consumer experience".

F10 describes digital health technologies as technologies used to deliver healthcare services remotely or to enhance in-person care, and gives telehealth, mobile health apps, patient portals, remote monitoring devices and secure messaging platforms as examples. Whether a particular system falls within F10 is a question for the RACGP or your own adviser. For tools outside that description, its steps are still a sensible checklist.

Do we need to pilot it first?

For anything that changes how the team works, yes: try it with a few people before everyone. A pilot shows you what the vendor's demonstration did not, such as how the tool fits your appointment book, your forms or your support workers' shifts. Keep the old way running alongside it so that nothing is missed if the pilot stops.

For AI scribes, the RACGP's AI scribes page suggests, when assessing a scribe's usability, taking advantage of any free trial periods to test how the scribe works. Its other usability points include the time required to set it up, how it integrates with current clinical information and workflows, and whether training and ongoing support are available.

How do we train staff on a new system?

Before it goes live, on your own workflows, and again a few weeks in. Start with the people in the pilot, then have them help train the rest. A short session using your own appointment types, forms or shift patterns is usually more useful than a general product tour.

The guidance points the same way:

  • AI scribes. The RACGP says facilitating initial and ongoing education and training for all AI scribe users is important in helping to ensure proper use, and that education and training are often available through AI scribe vendors. It also says GPs might require training in the use of an AI scribe before officially implementing the technology into consultations, and that this would require an investment of time and money for practices.
  • Digital health technologies in general practice. F10 suggests the practice could have technical experts provide training tailored to the practice's needs and patient population, and involve them in workflow design when introducing new digital tools.
  • National digital health systems. The Australian Digital Health Agency's training and support page describes "Accredited professional development and education on a range of digital health initiatives". It links training on My Health Record, electronic prescribing and cyber security, webinars for healthcare providers and online training courses. Its Online Learning Portal has courses for everyone, for healthcare providers, for students and for professionals. Use these for the national systems your team works with.

How do we help support workers with low digital confidence use the app?

Give them time, practice and a person to ask. Support workers may be using a new rostering or notes app on their own phone, away from the office, with no one beside them. Before go-live, run a short hands-on session where each person logs in, clocks on and writes a practice note on their own device. Pair less confident workers with a colleague who is comfortable with the app for their first few shifts, and give everyone one phone number to call when they are stuck.

Keep the paper or old process available for a short, agreed period, and say when it ends. Ask at the two-week check-in which steps people are avoiding or doing twice: those are the places to retrain.

What should we look at in the two-week check-in?

Whether the tool does the job, and what it costs the team to use it. Ask the pilot users what took longer than expected, what they worked around, and what questions they could not answer. Check the work the tool produced: notes, recalls, rosters or letters. Then decide to go ahead, change something first, or stop.

Keep reviewing after go-live. F10 suggests the practice could review the use of the technology periodically to confirm it continues to support safe, high-quality patient care. For AI scribes, the RACGP's page says "The implementation of AI tools requires ongoing due diligence", and suggests monitoring the performance of any implemented scribe to ensure it remains fit for purpose.

How do we roll back if it is not working?

Plan the way back before you start. Write down what would make you stop, who decides, how you return to the old way, and what happens to the information already entered in the new tool. Ask the vendor before you sign how you can take a copy of your data and in what form.

A rollback is not a failure. It is the reason the pilot was small and the old way stayed running.

What about changing our main practice or NDIS software?

The same steps apply, with more at stake: the data move, the cut-over date and the vendor's support.

F10 is part of the RACGP's Standards for general practices, so it is not an NDIS provider's standard. For the wider picture, see the AI in healthcare teams hub.

This page is general information, not legal advice. It summarises RACGP, Australian Digital Health Agency and AGPAL material and does not say whether any tool, process or organisation meets F10 or any other standard. Confirm your obligations with the RACGP or your own adviser, such as your medical defence organisation or lawyer.

How we help: How it works

Sources

  1. RACGP, F10 – Digital health technologies, Standards for general practices (6th edition) (Undated)
  2. RACGP, Artificial intelligence (AI) scribes (Header 29 Jun 2026; publication info: first published Aug 2024, updated Oct 2025)
  3. Australian Digital Health Agency, Training and support (Last updated 14 May 2026)
  4. Australian Digital Health Agency, Online Learning Portal (Undated)
  5. AGPAL, Artificial intelligence in general practice: industry guidance for safe and smart use (08 Dec 2025)

A useful place to start

Better technology.
Better care, together.

Tell us about your team and what you want to make better. We’ll find a useful place to begin.

Start a conversation