PACES Buddy can provide a place to rehearse a history, organise a presentation and answer an examiner's questions between bedside sessions. It cannot establish that you can detect a physical sign or examine a real patient competently. The useful decision is therefore not whether an AI examiner replaces clinical practice, but which part of your next practice session it should prepare you for.
As checked on 26 September 2026, PACES Buddy describes itself as a free, open-source, non-commercial practice partner offering a simulated patient, an AI examiner and a marksheet. This article is published by iatroX and includes iatroX among the complementary learning options considered. A free alternative deserves assessment on its actual usefulness, not dismissal because it has a different business model.
What does free mean in PACES Buddy?
The project's about page, reviewed on 26 September 2026, describes managed access and an alternative route using your own Claude API key. These are different cost arrangements. Under managed access, the project supports the service subject to its operating arrangements. With your own key, charges from the model provider are separate from the project's price.
The privacy and access information, checked on the same date, describes fair-use limits for managed access and storage arrangements for signed-in practice history. Free access should not be interpreted as a contractual promise of unlimited capacity indefinitely. Equally, a separate API charge does not mean the project is secretly selling a paid course: it means the learner is choosing a different route to model access.
Before starting a revision block, establish which route you are using, what usage limits apply and where your practice history will be retained. Use fictional cases or properly de-identified learning material. Do not submit a patient's identifiable history to make a simulation feel more realistic, and do not upload confidential examination material.
What the interface can represent
The public product description, checked on 26 September 2026, supports an educational interaction with a patient and examiner, including typed exchanges and available dictation functions. Describing an examination manoeuvre and receiving a finding is a representation of that manoeuvre, not observation of your technique. PACES Buddy's product information is the basis for these feature descriptions; this article does not report a controlled performance test.
That distinction changes how you should use the result. A learner may ask an appropriate question and then fail to listen to a real patient's hesitant answer. Another may name every component of an examination but struggle with positioning, sequence or interpretation at the bedside. A text-based system has evidence of the words submitted, not necessarily of the clinical behaviour those words describe.
Use the simulator to make your reasoning explicit. Ask yourself what information you still need, which finding would change your differential and how you would explain uncertainty. Reserve claims about observed physical performance for someone who actually watched it.
Read the seven-skill marksheet as an observation problem
The Federation's PACES format guidance, checked on 26 September 2026, identifies seven assessed skills. The table below is an interpretation of what different practice methods can observe. It is not an official classification of PACES Buddy or a validation of its marksheet.
| Official PACES skill | What a conversational rehearsal can practise | What still needs direct human observation |
|---|---|---|
| Physical Examination | Explaining sequence and the purpose of a manoeuvre | Technique, positioning and a complete examination |
| Identifying Physical Signs | Interpreting a finding supplied in the scenario | Detecting and distinguishing signs in a real patient |
| Clinical Communication | History structure, questions and spoken explanations | Interaction with a real person, including non-verbal responses |
| Differential Diagnosis | Generating and defending plausible alternatives | Whether the reasoning rests on findings actually obtained |
| Clinical Judgement | Explaining priorities, uncertainty and proposed next steps | Judgement demonstrated in the full observed encounter |
| Managing Patients' Concerns | Asking about concerns and responding to simulated objections | Whether the response addresses a real patient's priorities |
| Maintaining Patient Welfare | Verbalising consent, comfort and respectful practice | Actual conduct, dignity, discomfort and safe examination |
A high mark for a verbally represented physical examination should therefore prompt a question: what did the system observe that justifies this judgement? It may have assessed a well-organised description. That is useful, but narrower than performing the examination.
The same principle applies to negative feedback. When a marksheet says that you failed to address a concern, identify the relevant exchange. Did you miss the concern, answer it indirectly or respond in words the system did not recognise? These are different problems and require different practice.
A bedside-to-simulation practice cycle
Consider a fictional trainee who has just completed a supervised cardiovascular examination. Their supervisor identifies a specific weakness: the findings were collected competently, but the presentation mixed observations, interpretation and speculation. The trainee does not need another evening devoted to reciting the entire examination sequence. They need to practise separating those layers.
A useful cycle starts with a short, de-identified learning objective: present the positive and relevant negative findings first, state the leading interpretation second, and explain the remaining uncertainty third. The trainee then chooses a suitable simulated encounter or rehearses that structure with a study partner. The aim is not to reproduce the patient they saw, but to apply the presentation skill to a different scenario.
During the first attempt, they complete the presentation without assistance. During the debrief, they identify the first sentence where interpretation was presented as an observed fact. They rewrite that sentence, then repeat the task with a changed case. The final step is another supervised bedside presentation, asking the observer specifically whether the distinction has improved.
This proposed cycle produces several kinds of evidence: what happened at the bedside, what the learner could articulate in rehearsal and whether the change transferred back to observed practice. None alone establishes readiness for the whole examination.
How to challenge feedback without simply rejecting it
Treat an AI marksheet as a set of claims about your attempt. For each important claim, record the behaviour alleged, the exchange supporting it and the change you intend to make. This is more informative than retaining only the total score.
Suppose feedback says your differential was too narrow. The useful response is not immediately to request a longer list. Identify which alternative was missing, whether the case actually supported it and what additional finding would discriminate between the possibilities. Otherwise, the correction may teach indiscriminate listing rather than clinical reasoning.
When the feedback appears wrong, preserve the relevant non-identifiable exchange and ask a supervisor or experienced peer to adjudicate the specific point. Repeated disagreement about the same skill may reveal a limitation of the simulator, an ambiguous case or a genuine habit in your own performance. A single favourable or unfavourable result does not resolve which explanation is correct.
Do not convert practice scores into a probability of passing. The official examination uses observed encounters and its own assessment arrangements, described in the Federation's guidance. An educational marksheet is not an independently validated substitute for that process.
Where iatroX adds a different practice structure
Per iatroX product information, September 2026, iatroX Simulations offers voice and text practice, a pause-and-coach practice mode and an uninterrupted exam mode. Its feedback is linked to the transcript by domain, with Tutor-led remediation and a prescribed next case. These are design features for structuring practice, not evidence that simulated performance predicts examination results. The current tracks can be checked through iatroX's examination catalogue.
The same September 2026 product information describes one complete simulation available free. The paid subscription combines question banks, Socratic Tutor, study planning, simulations and CPD tools for £99 paid upfront annually, equivalent to £8.25 a month billed annually, or £29 a month. Simulations are included, not a separate add-on.
That is a different proposition from a free PACES-specific project. The relevant value question is whether the connected practice, explanation and follow-up methods help with your PACES preparation, not whether you receive access to examinations you will never sit.
Which arrangement fits your situation?
For a candidate who already has reliable bedside teaching and mainly needs extra verbal rehearsal, PACES Buddy's free managed route may be sufficient, subject to its current access terms. There is no reason to buy another platform solely because a paid alternative exists.
For a candidate whose difficulty is turning feedback into a structured next attempt, compare the available remediation and progress workflows. A connected learning platform may be useful when it addresses that specific gap. Demonstrate the workflow before committing to a subscription.
For a candidate struggling to elicit signs, perform an examination or interact comfortably with real patients, prioritise supervised clinical practice. Neither a generous free allowance nor a sophisticated simulation dashboard changes what must be observed at the bedside.
Frequently asked questions
Is PACES Buddy completely free?
As checked on 26 September 2026, the project describes free managed access subject to its operating and fair-use arrangements, while using your own Claude API key can incur separate model-provider charges. Check the current access information before planning intensive use.
Can an AI PACES marksheet show that I am ready to pass?
Not by itself: it reports judgements about a simulated interaction, not the full range of performance observed in the official examination. Use it to identify questions for further practice and supervised assessment.
Should I use PACES Buddy instead of bedside teaching?
Use it between bedside sessions to rehearse history-taking, presentations and reasoning. Physical signs, examination technique and patient welfare still need appropriate real-world observation.