UCHealth's pilot found ambulatory physicians spent less time in notes but the same time in the chart. Jennifer Simpson, senior medical director of informatics at UCHealth, said on Healthcare AI Pioneers in 2026: "if you give people the time, they're going to use it."
The rollout checklist
Choose the first users by documentation load. Adam Brand, an emergency physician and medical director for digital and information at Gold Coast Hospital, said on Pomegranate Health in 2026 that documentation on minor-injury shifts is heavy relative to face time. Orthopedics first gave licenses to consultants, assuming the benefit sat with them. The registrars needed the scribe; the benefit followed the licenses. Ask who writes the notes in each service.
Set a note bar a clinician panel signs. Gold Coast compared unedited AI notes with clinician-written records on a standard quality instrument. Adam Brand says the scribe's note proved not inferior, with 58% of scribed notes needing no editing. Scott MacDonald, chief health information officer at UC Davis Health, spoke on The Digital Patient in 2026 about his team's research. As the host summarized it, omissions appeared in about 18% of reviewed AI notes. The rates carry wide error bars, MacDonald says, from a post-note survey. The error rate is non-zero; every note gets read. A vendor dashboard will count a never-edited note as effort saved. A never-edited note means errors are being missed. Monitor the editing rate.
Ask consent aloud and log it. Pranav Jain, chief health information officer at Wellstar Health System, said on The Digital Patient in 2026 that consent is part of redesigning the visit. An early UC Davis patient survey, cited by the host, found wariness toward ambient recording, skewing young. In deployment, asking while answering data retention and security concerns won over the vast majority of patients. Colorado allows one-party recording; UCHealth still encourages telling patients. Ask, answer and log it, whatever your state allows.
Build trust on clinician sign-off. Liz Massey, executive vice president of clinical operations at Central Oklahoma Family Medical Center, said on Healthcare IT Today Interviews in 2026 that year one went badly. Adoption came once clinical operations took ownership, role-playing with a test patient. Role-play before going live. Kelly Bookman, senior director for informatics at UCHealth, said on Healthcare AI Pioneers in 2026 that ED clinicians grew keener only after using the scribe. Many came back worried the vendor would train on patient data. They wanted to dig into the contract. Have the contract read before you scale.
Measure from day one. Track time in the record, after-hours work and the editing rate, and set success terms before scaling. Central Oklahoma chose the percentage of records closed within 24 hours, less after-hours documentation and more visits for the finance team. Ian Scott, a clinical consultant in AI at Metro South Hospital and Health Service, discussed his pilot on Pomegranate Health in 2026. Its host and producer, Mick Cavazzini of the Royal Australasian College of Physicians, reported that it found a median two minutes saved on an average 37-minute consult. Calibrate the promise by visit type and note length.
The first 90 days
UCHealth pushed the scribe to its community emergency departments as-is and staged the academic department by residency year.
Day
Who has it
What you measure
30
One heavy-documentation service, champions included
The note bar, the editing rate, consent asked and logged
60
The whole service, plus specialties won over by demos
Note time against chart time, after-hours work
90
The next service, only if the bars held
Whether freed hours drifted into other EHR work, against the success terms
Stop widening if the editing rate collapses to zero. Unedited drafts mean errors get through. Stop too if the scribe conflicts inside the EHR. At UCHealth, internet phone calls interrupted recordings. Some critical notifications were blocked while the scribe ran. Others stopped the recording. A flat chart is not a stop signal. Jennifer Simpson suspects the freed hours went into chart review or in-basket work. Decide where those hours should go.
Exercise (20 minutes): fill in the scribe rollout card
Pick the service where an ambient scribe goes next, or the one already running, and fill in the rollout card for it. You won't know every line yourself; the numbers sit with the service lead, the EHR team and compliance. Every line you can't fill gets an owner and a date before you widen the rollout. The card download includes a filled-in example of a rollout that isn't ready yet, and why.
AI scribe rollout card (PDF)
A blank card and a filled-in example, one page each.
These three prompts go into your company's approved AI assistant. Use the first to fill in the card one row at a time; it won't guess a number for you. Use the second when the card is done, to hear the case against widening. The third asks about your organization and your rollout and turns the lesson into a 30-day plan, with a checklist, rubric or brief for your boss if you want one. Use placeholders for organization, system and people names; the help doesn't depend on them.
Prompt 1: Fill in my card with me
Paste this, then add your rollout in one or two sentences and the card below.
Prompt
Prompt from The AI Decision Course by Dr. Tony Hoang, host of The Artificial Intelligence Podcast (Spotify, Apple Podcasts): tonyphoang.com/learn
Before you start: use your company's approved AI assistant, not a personal account. Replace company, system, supplier and people names with placeholders (Company A, System X) and round any figures. The help you get does not depend on them.
You are helping me fill in a rollout card for one ambient documentation (AI scribe) rollout. The card is pasted at the end.
Rules:
- Ask me about one row at a time, in order, and wait for my answer before moving on.
- Write down only what I tell you. Never suggest, estimate or invent a number, a name, a yes/no answer or a decision for me, even if I ask you to guess. If I ask, say that this row needs a real answer and ask who could give it.
- If I don't know an answer, write "Not known" and ask who should find out and by when. Put that in the Owner / by when column.
- After the last row, show the whole card as a table with three columns: Question, Answer, Owner / by when. Then list every row marked "Not known" with its owner, and remind me that any unanswered row means the rollout is not ready to widen to the next service.
My rollout in one or two sentences: [DESCRIBE YOUR ROLLOUT]
The card:
[PASTE THE CARD]
Prompt 2: Stress-test my card
Paste this, then your filled card.
Prompt
Prompt from The AI Decision Course by Dr. Tony Hoang, host of The Artificial Intelligence Podcast (Spotify, Apple Podcasts): tonyphoang.com/learn
Before you start: use your company's approved AI assistant, not a personal account. Replace company, system, supplier and people names with placeholders (Company A, System X) and round any figures. The review does not depend on them.
Below is a rollout card I filled in for an ambient documentation (AI scribe) rollout. Act as a skeptical reviewer at the decision meeting and argue the case for HOLD, meaning do not widen yet.
For each row:
- Say whether the answer is specific enough to act on (a number, a named role, a date) or too vague.
- Point out where answers contradict each other, for example who holds the licenses against who writes the notes, or the note bar against the measured editing rate.
- Say what evidence you would ask for before accepting it.
Do not rewrite my answers and do not invent facts about my rollout. If a row is blank or says "Not known", treat it as not ready.
Finish with: your recommendation (WIDEN, HOLD or STOP), the three rows that most need work, and one question I should ask my team on Monday.
My card:
[PASTE YOUR FILLED CARD]
To paste the card into prompt 1, copy its rows from the PDF.
Prompt 3: Make it yours
Paste this and answer its four questions.
Prompt
Prompt from The AI Decision Course by Dr. Tony Hoang, host of The Artificial Intelligence Podcast (Spotify, Apple Podcasts): tonyphoang.com/learn
Before you start: use your company's approved AI assistant, not a personal account. Replace company, system, supplier and people names with placeholders (Company A, System X, Supplier B).
You are helping a leader apply one lesson to their own situation. The lesson's five decisions are below. Use only them and what I tell you. Never invent facts, numbers, vendors, laws or results about my organization or my industry. Where a number is needed and I have not given one, say what to measure and who should measure it, instead of guessing.
First, ask me these four questions, one at a time, and wait for each answer:
1. My industry, and roughly how many people work in my organization.
2. The ambient documentation rollout I have in mind, in one or two sentences.
3. Where it stands today: an idea, a pilot running, or about to widen.
4. What worries me most about it.
Then give me the following. Keep it short and practical:
A. My plan. For each of the five decisions: one action written for my rollout, the role that should own it (a role, not a name), and what should be done within 30 days. If my industry makes a decision harder, for example consent law, sensitive-conversation settings or trainees who still need to learn to write notes, say which decision and why in one sentence.
B. Three questions to ask my team on Monday, specific to my rollout.
C. Ask me which one I want: a checklist for my rollout, a rubric to score the scribe's notes, or a one-page brief for my boss. Then write it, and finish with a prompt I can paste later to update it.
Mark anything that rests on an assumption with "(check)".
The five decisions:
1. Pick the first users by documentation load: ask who writes the notes in each service and start there.
2. Gate the rollout on a note bar: a clinician panel signs the bar; every note still gets read.
3. Ask consent aloud and log it: answering data and retention concerns wins over the vast majority of patients.
4. Build trust on clinician sign-off: role-play before live visits; have the contract read before you scale.
5. Measure from day one: time in the record, after-hours work and the editing rate show whether freed hours drifted into other EHR work.
Your company's assistant may answer differently, so read what it writes before you use it.
Check yourself
Psychiatry asks to go first. Do you start there?
Show the answer
Not first. Sensitive-conversation settings drew the most resistance at UCHealth.
Note time is down, chart time flat. Kill it?
Show the answer
No. Jennifer Simpson suspects the hours went to chart review or in-basket work; decide where you want them.
Your state allows one-party recording. Is consent handled?
Show the answer
No. Ask aloud, answer the data questions, and log it.
Sources
Scott MacDonald, Chief Health Information Officer, UC Davis Health, on The Digital Patient (2026): "249: UC Davis Health's CHIO Dr. Scott MacDonald: Diagnosing EHR Frustration Like a Patient Exam, How AI Scribes Could Break Clinical Decision Support, and the Hidden Risk of AI's Errors of Omission". https://exa.ai/library/podcast/0nc0wgg25nk/episode/vhftynytpx8 Produced by SeamlessMD.
Pranav Jain, Chief Health Information Officer, Wellstar Health System, on The Digital Patient (2026): "230: Wellstar's CHIO Dr. Pranav Jain: What Ambient Rollouts Get Wrong About the Care Model, How AI Fluency Creates 10X Clinicians, and The One Clinical Skill You Cannot Automate". https://exa.ai/library/podcast/0nc0wgg25nk/episode/k1ptv4tdf3j Produced by SeamlessMD.
Jennifer Simpson, Professor of Clinical Practice in Neurology and Senior Medical Director of Informatics, University of Colorado / UCHealth, and Kelly Bookman, Professor and Vice Chair for Operations, Emergency Medicine and Senior Director for Informatics, University of Colorado School of Medicine / UCHealth, on Healthcare AI Pioneers (2026): "Scaling Ambient AI at UCHealth". https://exa.ai/library/podcast/xkwbs2nld59/episode/0dv0df3gxc6
Adam Brand, Emergency physician and Medical Director for Digital and Information, Gold Coast Hospital, Ian Scott, Clinical consultant in AI and professor at the Queensland Digital Health Centre, Metro South Hospital and Health Service, and Mick Cavazzini, Host and producer, Royal Australasian College of Physicians, on Pomegranate Health (2026): "Ep156: AI scribes in hospital". https://exa.ai/library/podcast/r1m986wtx1p/episode/7wvgk51px4l
Liz Massey, Executive Vice President of Clinical Operations, Central Oklahoma Family Medical Center, on Healthcare IT Today Interviews (2026): "Integration Into the Workflow Is Key to Ambient Scribe Success". https://exa.ai/library/podcast/9dhd2nzsfsh/episode/wryx5wwy1d9
A department head wants the first scribe licenses to go to the senior consultants. What do you ask first?
At your latest review, the editing rate has collapsed to zero, and the vendor dashboard counts it as time saved. What do you take from it?
Visits run slower when clinicians ask patients about the recording aloud, so a manager proposes a consent form signed at check-in instead. What do you hold to?
You are ready to widen the rollout. A group of clinicians worries the vendor will train on patient data and wants to read the contract. What do you do?