Before, during, after: the complete map of a patient visit — and what AI can take off each stage

A patient visit isn’t one event. It’s three.

There’s the work before the patient walks in — knowing why they’re coming, gathering the history, surfacing the red flags worth your attention. There’s the visit itself — the conversation, the exam, the decision. And there’s the work after — the note, the orders, the follow-up message, the loop that has to close before you can move on.

Almost every AI tool sold to physicians today aims at exactly one of those three stages: the middle. An ambient scribe listens to the conversation in the room and turns it into a note. That’s genuinely useful — but it’s a map with two-thirds of the territory missing. And the research is now clear that the middle is the smallest of the three levers.

Here’s the whole map, stage by stage, and what AI can actually take off each one.

Stage 1 — Before the visit: the most underused lever in primary care

By the time most physicians sit down with a patient, the history is still ahead of them. Chief complaint, history of present illness, review of systems, medications, allergies, relevant past history — all of it gets gathered live, in a 15-minute slot, while the clock runs.

This is the stage with the most slack in it, and the evidence is striking. In a controlled study of structured pre-visit intake, researchers had clinicians write a note three ways: from scratch, with an educational aid, and starting from a smart pre-visit intake form the patient completed beforehand. Writing from scratch took about 17 minutes per encounter. Starting from the pre-filled intake brought it to 5.7 minutes — a 2.98× speed-up — with note quality holding or improving on comprehensiveness and pertinent positives/negatives. ¹

That’s the whole game in one finding: when the structured history already exists before the visit, everything downstream gets shorter and better. The conversation starts from “let’s talk about what’s going on” instead of “let me get your basics.” Yet pre-visit intake is the stage almost no scribe touches, because a scribe can only record what’s said in the room — it can’t do the gathering that should have happened before the room.

This is also the stage where red flags can surface early. A structured intake that asks the right follow-ups can flag the chest-pain-with-exertion or the unintentional weight loss before you walk in, instead of hoping it comes up in the time you have.

Stage 2 — During the visit: real, but smaller than it’s sold as

The in-room stage is where ambient scribes live, and it’s worth being precise about what they do and don’t change — because the recent, well-designed studies agree, and the answer surprises people.

A 2026 npj Digital Medicine before-and-after study of 12 GPs across 535 observed consultations found an ambient scribe cut documentation time by 42.7 seconds per consultation — a real win — but total consultation time didn’t change. ² A 2026 JMIR Medical Informatics time-motion study of 169 consultations in Singapore found the same shape: documentation time down 15%, eye contact up 10.6%, and no change in consultation duration or total time per patient. Its conclusion is the line every physician should sit with: ambient scribes “reallocate clinician effort toward patient interaction rather than enabling faster patient turnover.” ³

Read that carefully. The in-room AI makes the visit better — more eye contact, less typing, lower cognitive load. What it does not do is shrink the visit or meaningfully shorten the day on its own, because the conversation still takes as long as a conversation takes. The middle stage was never where most of the recoverable time was hiding. It’s where the quality lives, not the hours.

Stage 3 — After the visit: closing the loop

Then comes the tail of the visit: finishing the note in your format, queuing the orders, drafting the after-visit summary, and handling the message that arrives two days later when the patient has a question. This is the stage that quietly stacks up — the part that, across primary care, pushes documentation into the evening (we mapped exactly where those hours go in the real anatomy of a physician’s day).

AI helps most here when the work of the first two stages flows into it — when the structured intake from before and the conversation from during arrive already organized, so the after-visit note is a review-and-sign, not a write-from-scratch. The further upstream the structure starts, the less there is to clean up downstream.

Why the whole map matters: the stages compound

Here’s the part that the single-stage tools miss. The three stages aren’t independent — they compound.

A great pre-visit intake (Stage 1) makes the conversation more focused (Stage 2), which makes the note faster to finalize (Stage 3). Conversely, a tool that only optimizes the middle inherits an ungathered history on one side and an unstructured note on the other — so its gains stay capped at “less typing in the room.” That’s exactly why the ambient studies show better visits but unchanged total time. They’re improving one link in a three-link chain.

This is the difference between an AI medical scribe and an AI clinical partner.

• A scribe is a microphone for Stage 2. It transcribes what was said.
• A clinical partner works all three stages: it runs the pre-visit intake and history (before), organizes the clinical data so the conversation stays on the medicine (during), and drafts the note for your review (after).

Claire is built as the second thing — an AI senior resident, not a recorder. It takes the history before you walk in, structures it for the visit, and drafts the documentation after, so the gains compound across the whole map instead of stopping at the room door. And the line that governs all three stages stays fixed: Claire drafts, you decide. Every note is yours to review, edit, and sign. The judgment never moves.

The visit was always three stages. The only question is whether your tools cover the whole map — or just the middle.

Want to see how Claire works before, during, and after the visit? See how Claire works, or book a demo.