Healthtech, news, and other cool things. We help over 30k providers get home early.
🇨🇦
Joined June 2023
- Tweets466
- Following197
- Followers69
- Likes1.1K
Pinned Tweet
Family doctors spend ~19 hours/wk on admin work, Scribeberry saves them ~2 hours/day
blog.scribeberry.com/scribeb…
2
8
3,748
Ambient scribes don't delete documentation work. They move it, from typing in the room to remembering afterward what the note got wrong.
17 Toronto family physicians, @chai_toronto, JMIR Med Inform, Nov 4 2026. We build these, so worth saying plainly.
10
Your saved CPSO policy links 404 now. @cpso_ca swapped policies for Professional Obligations.
New in Practice Management: only use an EMR provider willing to make records accessible to a CPSO investigation, and make sure they know it. That one's on vendors like us.
9
ED ENHANCED SHADOW-BILLING PREMIUM on your November RA is the 2024 PSA emergency department visit premium.
Second of two payments this year, for services Oct 1 2025 to Mar 31 2026. EDAFA workload model and fee for service only ED sites.
OHIP 261101, @OntariosDoctors: ontario.ca/document/ohip-inf…
17
Three hours longer in the ED. More admissions. Every number a dashboard watches got worse.
Those patients spent 7.5 more days alive and at home that year.
We measure the fast things because they're easy to measure.
Acad Emerg Med, Nov 1 2026.
13
BC dropped the prior auth form for first line COPD inhalers in 2020. Starting on the right one rose 34.8%.
Total drug costs didn't change, so the form was never rationing spend. It was just picking the first prescription.
@journal_CHEST, Nov 5 2026.
13
Every EMR we've worked in has a mandatory dropdown whose only honest answer is Other, and a free text box underneath holding what actually happened.
Guess which one ends up in the quality report.
14
A scribe hears what you said, not what you noticed and never said out loud.
New JGIM perspective calls that an epistemological gap, and asks health systems to judge scribes on reasoning fidelity, not transcription accuracy. Ours included.
link.springer.com/article/10…
7
An AI scribe gives you an edit button. That's control over one note.
It isn't control over the system that writes the next hundred the same way.
A new Frontiers in Digital Health paper calls that gap oversight asymmetry. Ours has the button too.
pmc.ncbi.nlm.nih.gov/article…
18
Most Colleges never say whether an AI scribe's audio and raw transcript belong in the chart. Quebec's does: draft aids, destroy them once the note is final.
Everywhere else the vendor's retention setting decides it by default. Ours too.
@CMPAmembers AI Scribes FAQ
11
Bill C-36 would move oversight off @PrivacyPrivee and leave anonymized data unregulated. That's what AI gets trained on.
Deliberately reidentifying it is an offence. Using it is governed by nothing.
Vendors like us benefit from that gap. We shouldn't.
pmc.ncbi.nlm.nih.gov/article…
13
Alberta Health Services built an in house AI scribe instead of buying one.
197 ED physicians, 21,548 sessions, under US$30 per physician per month against commercial US$99 to $600. Ours included.
Physicians rated note quality lowest of every item.
pmc.ncbi.nlm.nih.gov/article…
12
Radiologists told Parliament this month that the AI they want isn't a faster reader. It's decision support at the ordering end, so the right scan gets booked once.
We build documentation tools. The useful AI here sits upstream of us.
Canadian Assoc of Radiologists, HESA, Nov 3.
12
33% of referrals arrive incomplete or incorrect. 10% never send. Clinics take about 9 calls a day from patients asking where theirs went.
The referral limbo @CDHughett described, measured. BC's Digital Referrals and Orders, Nov 4 2026, 159 onboarded sites.
1
1
17
The College of Physicians and Surgeons of BC is retiring its four hour Medical Record Keeping course for family physicians at the end of 2026, after nearly two decades. The 101 and 201 online modules stay.
We build the thing writing the notes. The timing isn't lost on us.
14
Patients agreed they should be told when AI is used in their care, 4.48 of 5. On opting out, neutral at 3.33.
We spend more time on the opt out than on the telling.
PLOS Digital Health Nov 4. 20 US patient advisers.
pmc.ncbi.nlm.nih.gov/article…
9
A good AI note reads like prose. Coding, reporting and research run on discrete fields.
Waterloo Regional Health Network, now organization wide with AI scribe, says it's tracking exactly that. We don't see it measured often.
wrhn.ca/news-stories/ai-scri…
15
Ambient AI is permitted for students in 3% of US family medicine clerkships. Of the directors opposed, 85% worry about communication skills and 82.5% about clinical reasoning. @STFM_FM, Nov 2.
We build the thing that writes the note, and that's still a fair worry.
1
17
Same day note closure rose 10.2 points with an AI scribe when the clinician spoke the patient's language. Add an interpreter and the gain was inconclusive. @JAMANetworkOpen, Nov 6. Ours hasn't solved that room either.
pmc.ncbi.nlm.nih.gov/article…
14
The 10% to 30% documentation time saving in AI scribe pitches, ours included, rests on a thin base.
New systematic review: 520 records screened, 7 studies included. All US, mostly observational, short follow up.
pmc.ncbi.nlm.nih.gov/article…
11