ClinicFrame
Like Granola, but for healthcare. Fully HIPAA-compliant.
Start new thread

ClinicFrame - Like Granola, but for healthcare. Fully HIPAA-compliant.

by•
An ambient AI scribe that writes clinical notes in real time so you can focus on your patient. It passively captures every visit, whether in person or virtual, and delivers a complete, structured clinical note the moment the encounter ends. HIPAA compliant, desktop native, and EHR ready in seconds. We're launching Scribe today, with a much bigger vision ahead: a medical intelligence platform that goes beyond documentation to support the entire clinical workflow.

Add a comment

Replies

Best

I spend a lot of time looking at AI applications in regulated industries, and most of them die in the gap between the demo and the compliance review. Starting from compliance and working toward the delight is the harder order to do it in, and the one that survives.

Congrats to the team. Rooting for this one.

Maker

 
Thanks. That gap is real, and the reason the order matters is that compliance can't be retrofitted. You can add polish to a compliant product. You can't add a BAA to a data path that was built without one.

The question that kills those demos is never about the model. It's where the data sits and who signs for it.

Not in healthcare myself, but I have watched this team work and the discipline is the thing. Narrow problem, real users, no attempt to be everything at once. Congrats on shipping. Curious how you think about expanding beyond the note itself once clinicians trust the workflow

Maker

 

Thanks. The expansion follows the note, it doesn't leave it.

Trust gets earned in order. First the note has to be good enough that you stop editing it. Then the patient context between visits, because the note already holds it and rebuilding that history by hand is the next tax after documentation. Then the revenue side, because the billing code comes out of that same note.

Each step is only allowed by the one before it. We don't get to touch a claim until the note is boring and reliable.

super needed!!! how do you find yourself different to fireflies?

Maker

 
Thanks. Fireflies is built for meetings, we're built for visits, and that changes almost everything downstream.

Three concrete ones. Nothing joins your call as a third participant, which matters because plenty of visits happen in a room, and a bot sitting in a therapy session is a non starter. The output isn't a summary with action items, it's a structured clinical note in the format your EHR and your payer expect. And the unit of memory is the patient over time, not the meeting.

Compliance is the floor: a BAA comes with the account at self-serve price, not as an enterprise add-on.

Ambient audio capture of clinical encounters is genuinely useful but it's also the highest-stakes category for AI hallucination, a fabricated medication dosage or an incorrectly captured allergy in a clinical note is a patient safety issue, not just a bad output. What's the accuracy validation methodology for ClinicFrame's notes, like how are you measuring and communicating accuracy to clinicians who are relying on these notes for clinical decisions?

Maker

 

You're separating the right things. A missed nuance in the narrative and a fabricated dosage are not the same class of error, and reporting one accuracy number is how that second one stays hidden.

So we don't track a single score. We look at error rates by content class, and we count omissions and fabrications separately, because they have opposite fixes. For the high-risk classes (medication, dose, allergies, any number) the rule is that nothing enters the note unless it was said in the encounter. The model doesn't complete from prior knowledge, and anything unclear is marked as missing rather than filled in.

The second half of your question matters more than the first. Every line traces back to the audio it came from, so review takes seconds instead of a full re-read, and the note stays a draft until the clinician signs it. Accuracy a clinician can't inspect isn't accuracy, it's a claim.

If you've done validation work in this space, I'd want that conversation.

Great & Genuine Concept , don't you think there's need of mobile app for Clinic Frame, To personalize experience?

Maker

 

Hey, this is a wonderful idea and we are just waiting for the app store approval right now!

Not a clinician. I build compliance software, so I'm answering sideways.

The one thing I'd never let a tool do: change anything after a human signed it. Silently, in a background job, with no record of what moved.

Every AI action in my product needs an explicit confirm and an audit log entry. Because "who changed this and when" deserves a better answer than "the model, sometime Tuesday."

A signed clinical note is a legal document. An AI that quietly edits it after the fact hasn't improved it. It's forged it.

You've clearly thought about this already. The part I'd stress-test is post-signature: re-runs, template migrations, retro-enhancement of old notes.

Congrats on the launch.

Maker

 

You're not answering sideways, you're answering the part most people skip.

Acceptance closes the document on our side (the formal signature usually lives in the EHR). After that, nothing is edited in place. A change becomes a new versioned entry with its own author and timestamp, and the original stays exactly as it was accepted. That's the amendment model clinical records have used for decades, and there's no reason for AI to get an exception.

On your three: a better model never re-runs an accepted note, it only touches drafts. A template change applies to future notes and never reformats past ones, so a note keeps the shape it had the day it was accepted. Retro-enhancement is the one we rule out entirely, and it's the hardest of the three, because it never arrives as an attack. It arrives as a well-meaning product suggestion: we could re-run last year's notes with the new model and they'd all be better.

"Forged it" is the right word for that. If you ever want to compare audit trail designs, I'm in.

the shadow-IT framing is the strongest part here tbh, you're not selling a new habit, just a compliant version of one clinicians already formed on their own. the thing I'd want to know is edit friction, since the clinician signs the note and carries the liability, so time-to-correct probably matters more than aggregate accuracy.

Maker

 

Right, the habit was already there. The only thing we change is where the data goes.

On edit friction, agreed, and the reason is that finding the error costs more than fixing it. Correcting two words takes seconds. Reading the whole note to be sure there's nothing to correct is the real tax, and that's the one nobody measures.

So what we watch is time from opening the note to accepting it, and how much of that time is reading versus typing. A note that needs no edits but three reads to trust is worse than one where you fix a word and move on. Tracing each line back to the audio exists for exactly that: you check the one line you doubt instead of re-reading the encounter.

 the read-versus-type split is a smart thing to be watching. one blind spot in it though: a fast accept is ambiguous. it is either trust you earned or someone not really reading, and both look identical in that metric.

the tell would be whether the notes accepted fastest are the ones amended later. if they are, that is complacency showing up rather than accuracy. line-level traceback back to the audio is the right fix for it either way, since it makes spot-checking cheap enough that people actually do it.

Hey Clemente, this looks really cool, congrats on the launch 🚀

You mention that it’s EHR-ready. Are you initially focusing on specific countries and healthcare systems, or is ClinicFrame designed to be international from day one? Healthcare regulations and EHR ecosystems vary so much between countries, so I’m curious how you’re approaching that.

Best of luck today!

Maker

 

Thanks. US first, and deliberately so.

Compliance doesn't generalize. HIPAA with a BAA is a specific regime, and PIPEDA or GDPR are not the same work with a different label on it. Doing two jurisdictions halfway is worse than doing one properly, especially when what you're selling is trust.

On EHR-ready it's worth separating two layers. Getting the note into a chart is the universal part: structured text in the format the record expects, which works in any system in any country. Direct integrations are the part that goes system by system, and those we do one at a time, starting with what our users are actually on.

Canada is the request we hear most after the US, so PIPEDA is the next jurisdiction we look at rather than a hypothetical one.

The versioned-amendment answer to Maciej and the traceable-evidence answer to Ansari are the two best answers I've seen on this kind of product, most scribes wave away the audit trail question. One thing I didn't see covered: a lot of visits aren't just doctor and patient, there's a caregiver, a family member translating, sometimes an interpreter on speakerphone. Does the note distinguish who actually said the symptom versus who's relaying it secondhand, or does that nuance flatten out in the transcript the same way it would in a rushed human note?

First
Previous
12