AI consultation practice has become a regular topic in GP trainee conversations. If you’re curious but haven’t tried it — or if you’re sceptical — here’s a straightforward account of what the experience is actually like, what it does well, and where it falls short.

What happens

You open a case — maybe a patient with worsening breathlessness, or someone presenting with low mood. A timer starts. You speak out loud as you would to any patient in surgery.

“Good morning, I’m one of the GPs here. How can I help you today?”

The AI patient responds with voice. They describe their symptoms, answer your questions, share concerns. If you ask about their family history, they’ll tell you. If you miss an important question, they won’t volunteer the information — just like a real patient.

The conversation goes back and forth in turns. You lead it, the patient follows your cues. When you’ve finished — either because you’ve closed the consultation or the timer runs out — you get a feedback report.

That’s genuinely it. No special equipment beyond a phone or laptop and a reasonably quiet room.

What the feedback tells you

This is where the value sits, and where platforms differ most.

At a minimum, you should get a breakdown across the three consultation domains: how well you gathered information, whether your management plan was appropriate, and how effectively you communicated. Without this, you’re practising in the dark.

Better feedback goes further — identifying specific moments. “You acknowledged the patient’s worry about their mother’s cancer but didn’t explore it further.” That’s actionable. “Your data gathering could be improved” is not.

The most useful feedback tracks your performance across multiple cases, so you can see patterns. Maybe you consistently rush safety-netting. Maybe your data gathering is strong with medical cases but thin with mental health presentations. These insights only emerge over time, and they’re the things that actually change how you consult.

What it does well

It’s there when your study partner isn’t. This is the honest reason most registrars use it. You want to practise, it’s 9pm, nobody’s available. AI practice fills that gap reliably.

The feedback is consistent. A study partner’s feedback varies depending on their experience and confidence. AI feedback applies the same criteria every time. That consistency is useful for tracking genuine improvement.

You can repeat without awkwardness. If you handle a case badly and want to try a different approach, the AI resets cleanly. Repeating a case with a human study partner who already knows the outcome doesn’t work the same way.

Where it falls short

Emotionally complex cases. Breaking bad news to someone who responds with shock or grief is fundamentally different when it’s a human sitting across from you. For the most interpersonally demanding cases — safeguarding, bereavement, the patient who makes you uncomfortable — practise with a real person if you can.

It doesn’t teach you the medicine. If your management plan is wrong, the feedback should flag it. But learning the correct guidelines is still your job. AI practice tests your consulting, not your clinical knowledge.

It’s not group learning. Watching a colleague consult and discussing it afterwards builds understanding in ways that solo practice can’t. AI practice sits alongside VTS teaching and group tutorials — it doesn’t replace them.

How to judge a platform

If you’re considering trying AI consultation practice, a few things to look for:

Is it voice-based? Text-based platforms don’t build the skill you need. You have to practise speaking.

Is the interaction turn-based (you speak, patient responds) or open-listening (the AI talks freely)? Turn-based is more realistic — real patients don’t monologue, and the purpose of data gathering practice is learning to elicit information through questions.

Can you practise without rationing? Credit or token-based models create anxiety about wasting attempts. Learning requires repetition, especially for cases you find difficult.

Is there a free trial that shows you the full feedback experience? You shouldn’t need to pay before you know whether the feedback is useful to you.

Does it work on mobile? Many registrars practise during downtime — between clinics, on a break, at home in the evening. A platform that requires a desktop limits when you can use it.

The honest place for it

AI consultation practice is a tool. It’s good for volume, consistency, and availability. It’s not a replacement for peer practice, trainer feedback, or clinical experience. The registrars who use it best treat it as their regular solo practice routine — two or three sessions a week — and reserve human interaction for the cases where it genuinely makes a difference.

---

We designed ConsultMentor as GP trainers who wanted something our own registrars could use between teaching sessions. Try a free case — it takes about 15 minutes and shows you what the feedback looks like.