“What would you do with 45 minutes back, three evenings a week?” 

That’s the question Healthbridge’s Heleen Fengler recently put to a room of doctors during a discussion about Ai & operational excellence. Her point was simple: clinical work doesn’t necessarily end when the last patient leaves. Notes, lab follow-up, recalls & other admin work can easily spill into evenings & weekends. 

That’s where the promise of Ai in a medical practice becomes interesting. Not in how clever the technology sounds, but in whether it can take some of the repetitive work off your plate, help you find the information you need & make everyday tasks easier.

Getting there, though, starts with knowing how to ask.

The good news is that you don’t need to become a prompt engineer. Often, the difference between a generic Ai response & something genuinely useful comes down to giving it a better brief & knowing what to ask next.

Give Ai a proper brief

Think about how you would brief a new assistant. If you simply said, “Write a letter”, they would have to make a lot of assumptions. Ai is much the same.

A useful prompt usually gives it four things:

Task + context + important details + desired output

Here’s what that looks like in a clinical setting:

Instead of asking…	Try asking…	What you’ve added.
Summarise this patient’s history.	Summarise this patient’s diabetes history over the past two years. Include HbA1c trends, current medication, relevant investigations & changes to the management plan.	Condition, timeframe & relevant information
Write a referral letter.	Write a referral to an endocrinologist for diabetes management. Include diagnoses, current medication, examination findings & recent investigations.	Audience, purpose & what to include
Explain these results.	Explain these blood results to the patient in simple English. Highlight what they need to know & what they should do next.	Audience, language & desired outcome
Write a medical report.	Write a confidential medical report for insurance purposes. Include diagnosis & current treatment only. Use prose, no bullet points, & keep it under four paragraphs.	Purpose, exclusions, format & length

The aim isn’t to make your prompts longer for the sake of it. It’s to give Ai the information it needs to understand what a useful result looks like.

These examples are the types of tasks Healthbridge Clinical Assistant can help with, including medical histories, referrals, visit summaries, motivation letters & custom clinical documents. 

Don’t try to write the perfect prompt

One of the easiest mistakes to make with Ai is assuming you need to fit everything into the first question. You don’t. Think of prompting as a conversation rather than a search. Start with a useful question, look at the response & decide where you want to go next.

For example, you might start with:

Summarise this patient’s diabetes history over the past two years. Include HbA1c trends, current medication & relevant investigations.

Once you have the summary, continue:

Now turn that into a referral letter to an endocrinologist. Include the current management plan & reason for referral.

Then refine it:

Make the letter more concise & put the most clinically relevant information first.

And finally:

Now create a patient-friendly summary explaining what we discussed & what the patient needs to do next. Use simple English.

You didn’t need to anticipate every step before you began. One conversation has moved from reviewing a patient’s history to preparing a specialist referral & creating information for the patient.

Build on what comes back

This becomes even more useful when the question doesn’t have a straightforward answer.

Imagine you’re concerned about bad debt in your practice. You could ask:

How do I reduce bad debt?

A general Ai tool could give you some useful suggestions. But when Ai can work with your practice data, the conversation can become much more specific. 

1. Explore: Analyse my payer mix & outstanding accounts & identify where my biggest bad-debt risks are. 
2. Investigate: Which of these risks appears to be contributing most to delayed or lost revenue?
3. Dig deeper: Look more closely at my rejected claims. What patterns can you identify by medical scheme, rejection reason or diagnosis coding?
4. Prioritise: Which three issues should I investigate first?
5. Act: Give me a practical action for each & suggest how I could monitor whether it improves.

The first question opens the door. Each follow-up narrows the problem until you have something you can investigate or act on. This is where a tool like Healthbridge Practice Agent becomes useful: you can ask questions of your own practice data in everyday language & then keep drilling into what you find. 

You can apply the same approach to other questions about your practice. If Ai identifies something interesting, ask it to break the finding down further. Ask what changed. Compare periods. Look for patterns. Ask which issue deserves attention first.

Sometimes the most useful prompt is simply the next one.

Ask Ai to turn information into something useful

Ai isn’t only useful for finding or summarising information. It can also help reshape information for what you need to do next.

The same clinical information, for example, might need to become a detailed referral for a specialist, a motivation for a medical scheme or a simple explanation for a patient. So tell Ai who the information is for & what you want them to get from it.

You can also refine the response without starting again:

• Make this shorter.
• Explain it in simpler language for the patient.
• Focus only on the past 12 months.
• Put the most clinically relevant information first.
• Remove information that isn’t relevant to the referral.
• Turn these findings into an action list.

This is where conversational prompting can save time. You’re not repeatedly recreating the work. You’re guiding the Ai towards the version you actually need.

Need some prompts to get started?

You don’t always have to start with a blank box.

Healthbridge’s Clinical Assistant Smart Prompt Guide includes 61 prompt templates covering medical history reports, motivation letters, referrals, visit summaries & custom documents. Use them as a starting point, then adapt the prompt for the patient, purpose & output you need.

DOWNLOAD THE CLINICAL ASSISTANT SMART PROMPT GUIDE

Ai can assist. You still decide.

Ai is there to assist, not replace your judgement. Just as you would review work prepared by an assistant, check Ai-generated clinical documents before you use them. Recommendations still require professional judgement, patient information needs to be handled appropriately, & anything being sent to a patient, specialist, medical scheme, employer or other third party should be reviewed before it leaves the practice.

The same principle runs through Healthbridge’s approach to clinical Ai : technology supports the work while the clinician retains authority over the final decision. Good prompting therefore isn’t just knowing what to ask Ai to do. It’s also knowing where its role ends & yours begins.

Put Ai to work in your practice

If you’ve used ChatGPT before, the experience of talking to Ai in your practice doesn’t need to feel unfamiliar. You ask a question in everyday language, see what comes back & keep the conversation going.

The difference is that Healthbridge’s Ai tools are built into the workflows & information you already use in your practice.

With Healthbridge Clinical Assistant, you can ask Ai to help you work with information already captured in a patient’s clinical record. Ask it to summarise a medical history, prepare a referral or motivation letter, or turn clinical information into a simple visit summary for the patient.

Healthbridge Practice Agent applies the same conversational approach to the business side of your practice. Instead of digging through reports to understand what’s happening, you can ask questions of your practice data in plain language. Start with “Where is my bad debt coming from?”, then keep going: “Which schemes account for most of it?” or “What should I investigate first?”

That’s really the point of better prompting. You don’t need special Ai language or a perfectly constructed question. Ask it the way you’d ask a capable assistant, then use the answer to decide what to ask next.

And because these tools sit within the Healthbridge practice environment, Ai becomes part of the work you’re already doing, not another standalone tool to manage. As Heleen put it: “Reclaimed time belongs to family, rest, or a career choice, not automatically to more patients.”

Want to see what Ai could look like in your practice? Healthbridge works with more than 7 000 private healthcare practitioners, helping them use technology to run smarter, more efficient practices. Speak to an experienced Healthbridge Business Consultant about how Ai could help you run your best practice.