The consultation has already started before the patient walks in
There is a moment most clinicians recognise. A patient arrives, you open their file, and you spend the first ninety seconds reading backwards through their history - scanning notes, checking what was prescribed last time, trying to recall whether that referral was ever followed up. The patient is watching. The clock is running.
It is not a memory failure. It is a structural one. The information was always there. The problem is that nothing prepared it for you.
Most of the conversation about AI in clinical practice centres on what happens after the consultation ends - the SOAP note generated from a transcript, the referral letter drafted from a recording. That is a real and useful capability. But it treats AI as a tool for documentation, and documentation as something that happens once the clinical work is done.
That framing misses a more significant opportunity.
What good preparation looks like
A clinician seeing twenty or thirty patients a day cannot hold all of them in working memory between appointments. That is not a failure of diligence - it is a cognitive reality. Each patient arrives with a history that may span years, multiple conditions, a stack of documents, medications, referrals, and correspondence. Reading all of it before every appointment is not realistic. Not reading any of it is worse.
What clinicians need before a consultation is not the complete record. It is a synthesis of the complete record - the relevant history surfaced, the flags visible, the trajectory of care made readable in the two minutes available between appointments.
That synthesis is exactly what AI is well-suited to produce. Not from a transcript, because there is no transcript yet. From the patient record that already exists.
The scribe model asks: what happened in this session? The pre-consult summary asks a more useful question - what does this clinician need to know before the session begins?
The prompt that changes how you start a consultation
In Bookem, AI Assist works by applying a prompt to whatever context you give it. Most clinicians use it after a session - they record a consultation, apply a prompt, and get a structured note back. That workflow is fast and it works.
But the prompt can be applied before the session too. The input is not a recording. It is the patient record: their history, past session notes, intake forms, active conditions, medications, clinical alerts, previous documents. Open the patient's profile, select your pre-consult summary prompt - or write one in seconds - and run it. AI Assist draws on the connected record and returns a structured summary oriented around what matters for today's appointment.
That summary might include a chronological overview of the patient's care, the key clinical context for this appointment type, flags worth noting, and open threads from the last session. It takes moments to generate and a minute to read.
The quality of that output depends entirely on the quality and completeness of the patient record behind it. A fragmented record - notes in one system, forms somewhere else, referrals in email - produces a shallow summary, or nothing useful at all. A complete, connected record produces something a clinician can actually use.
What changes in the room
The practical effect is a different quality of presence at the start of the consultation. Not because the clinician has spent more time preparing, but because the preparation happened in a fraction of the time it would otherwise require.
A GP opening a follow-up consultation already oriented to where things stood last time. A physiotherapist who walked in knowing which goals were set, what progress was reported, and what the patient said they were struggling with. A psychologist who reviewed the arc of the last few sessions before the patient sat down.
None of these are dramatic changes in isolation. Cumulatively, across a full day of consultations, the difference in cognitive load is significant - and the difference in the quality of care that load enables is real.
Why this only works inside an integrated system
A standalone AI scribe cannot produce a pre-consult summary. It has no access to the patient record - it only processes what you feed it in that session.
A pre-consult summary requires that everything relevant to the patient already exists in one connected place: intake forms completed before the appointment, past session notes structured and searchable, documents versioned and attached to the right profile, clinical alerts flagged and visible. When AI Assist draws on a Bookem patient record to generate a pre-consult summary, it is drawing on all of that simultaneously.
This is why the value of an integrated clinical system compounds over time. Every well-structured note, every completed intake form, every referral filed in the right place - these are not just documentation habits. They are the inputs that make future consultations better prepared.
The pre-consult summary is a new way of thinking about what AI in a clinical workflow is actually for.
Getting started
If you are already using AI Assist in Bookem, the pre-consult summary is available now. Before your next appointment, open the patient's profile and select your saved pre-consult summary prompt - or type one quickly - and run it. If you have not set one up yet, a prompt asking AI Assist to summarise the patient's history, flag anything clinically relevant to today's appointment, and surface any open threads from the last session is a practical starting point. Save it, and it is there for every patient from then on.
The first time you use it before a consultation you were not fully prepared for, the value is immediate.
Related reading
AI clinical documentation: why SOAP notes are just the starting point
The pre-consult summary is one expression of a broader argument: that AI in clinical practice becomes genuinely powerful when it has access to the complete patient record, not just the most recent session.
Want to see AI Assist working across the full consultation workflow?