Built on the floor. Not in a boardroom.By physicians who wanted better.
DAX answers to Microsoft. Abridge answers to its VCs. Scribeable answers to the physicians using it. Two-pass AI that asks before it guesses. Every sentence cited. Your notes, your account, your workflow — wherever medicine takes you.
No credit card required. 14-day trial with full Pro features.
236 calculators · 63 quality measures · 8,400+ HCC crosswalk · Two-pass verification
Recording Session
Internal Medicine Consult
Most scribes write what they heard. This one asks what it needs to know.
You speak. Forty-five seconds later, you have a note. Every scribe does that now.
What’s different: before the note finalizes, the system surfaces what’s clinically ambiguous or missing — and asks. Not a generic prompt. A specific question about this patient. Did you intend to document a Wells score? The HPI mentions chest pain — should the HEART score appear in the assessment? That’s the difference between a transcription tool and something that thinks alongside you in the room.
Stage 2 runs 236 clinical calculators against what you actually said — checks the math, corrects anything that drifted, and links every score to its clinical guideline. Every sentence in the finished note carries a citation: the exact transcript span it came from. Click any claim and see where it started.
No black box. If it’s in the note, you can see why it’s there.
±9% of your Medicare revenue is on the line every year.
First-generation ambient scribes contained hallucinated content in 31% of audited notes (Palm et al. 2025). The JAMIA DAX evaluation found “no quantifiable effect on patient safety and no benefit in productivity” (Haberle et al. 2024, JAMIA). Scribeable was engineered for the layer above the microphone.
We started with one physician.
Every feature that exists was tested in a real clinical situation before it shipped. No paid reviews. No marketing department. Physicians passing it to the next physician because it actually works the way medicine works.
1,900+
notes generated
across 46 specialties
0
data breaches
zero, ever, by design
0
dollars of outside capital
no VC pressure on the roadmap
Three things we built that most scribes skipped.
Not incremental improvements to transcription accuracy. Decisions about what the product should do that most scribes never made.
Five kinds of clinicians.
You haven’t heard these five complaints in one place. You’ve heard every one of them in your hospital cafeteria. Different role, different specialty, same vendor boardroom deciding how medicine gets documented this quarter.
The hospital-employed physician
“They're picking my tools without asking.”
The practice administrator
“They raise prices 40% at renewal. We have no leverage.”
The CMIO
“Our vendor got acquired, doubled pricing, and killed the features we actually asked for.”
The resident
“They're deciding how I'll practice medicine for the next 40 years.”
A different pitch for every buying unit
For Group Practices
One practice. One MIPS score. One billing line.
- •Practice-wide MIPS dashboard with shared templates
- •Seat math that works — self-serve or 20-minute group demo
For Health Systems
Choose the scribe your security review can actually pass.
- •AES-256-GCM per-org KMS keys, signed BAA on upstream LLM, 7-year audit retention
- •Parallel pilot alongside your current scribe — 30-day decision, no switching cost
Works with your EHR
Browser extension inserts notes directly into your EHR
See all integrationsWhy not just use Why not just use
Because neither will round with you tomorrow morning. Six specific things we built that a general-purpose AI can’t, and that the hospital-bundled scribes weren’t designed to.
Two-Stage AI That Asks Before It Guesses
Stage 1 drafts the note. If something clinically critical is missing or ambiguous, the AI asks you — not the other way around. Stage 2 verifies accuracy, re-validates calculator scores, extracts billing codes, and checks quality measures. The Palm 2025 audit found first-generation ambient scribes had hallucinated content in 31% of notes. Asking for clarification instead of inventing it is the architectural answer.
HCC V28 + MIPS Engine That Answers to You
8,400+ ICD-10 → HCC crosswalk entries, CMS V28 RAF calculation, 63 quality measures scored in real time, CPT-II codes auto-generated. ChatGPT won't do any of this — no BAA, no audit trail, no retention policy you can show a surveyor. DAX answers to Microsoft. Abridge answers to Lightspeed. Here the engine answers to the physician using it — and travels with you to every job you take.
Architecture You Can Actually Read
AES-256-GCM field-level encryption with per-organization keys (Google Cloud KMS). ECDH P-256 key exchange — plaintext keys never transmitted. Immutable INSERT-only audit trails. Your data never trains a model, anyone's. Every BAA we've executed is publicly listed at /transparency, including Anthropic. ChatGPT's terms forbid PHI entry entirely. Most enterprise scribes decline to publish their architecture at all.
Rounding Mode, Pre-Visit Briefs, 27 Note Types
One continuous recording during rounds auto-segments by patient with accumulated encounter history. Pre-visit briefs pull prior encounters, meds, labs, and allergies before you walk in. 27 note types across 46 specialties — each with specialty-specific prompts, billing codes, and quality measures. Built because a resident got tired of charting from memory at the end of a 6am rounds block. No general LLM ships this workflow. No enterprise scribe was designed for it.
Same Patient. Same Encounter.
See why physicians say Scribeable notes are “actually usable” — with integrated risk scores, billing codes, and clinical reasoning that typical AI scribes simply don’t generate.
HPI
62 y/o male presents with chest pain for 2 hours. Substernal, pressure-like, radiating to left arm. Associated with diaphoresis and shortness of breath. Pain started while climbing stairs. Patient has history of HTN, DM2, and hyperlipidemia. Takes lisinopril, metformin, and atorvastatin. Denies recent illness or trauma.
Assessment & Plan
*Note examples are illustrative representations based on common AI documentation patterns. “Typical AI Scribe” represents composite characteristics of standard tools and does not depict any specific product. Patient scenarios are entirely fictional. Clinical notes should always be reviewed by a licensed provider.
AI Medical Scribe Note Quality Comparison
Scribeable produces clinically enriched notes with integrated risk calculators (HEART Score, CHA₂DS₂-VASc, Wells PE, PHQ-9, Caprini VTE), automated billing code optimization (E&M levels, HCC/RAF capture, CPT codes), dangerous diagnosis exclusion documentation, evidence-based prescribing citations, CMS compliance for operative reports, HEDIS care gap tracking, and MIPS quality measure documentation. Typical AI scribes produce basic notes without these advanced clinical features.
ED Chest Pain: 62-year-old male presenting with substernal chest pressure, diaphoresis, and exertional dyspnea
HPI: 62 y/o male presents with 2 hours of substernal chest pressure, rated 8/10, radiating to left arm and jaw. Onset during exertion (climbing stairs). Associated with diaphoresis and exertional dyspnea. Pressure-like quality, unrelieved by rest. No pleuritic component, no positional variation, no reproducibility with palpation. Risk factors: HTN (10 years), DM2 (8 years), hyperlipidemia, 30-pack-year smoking history (quit 5 years ago), family history of MI (father at age 58). Current medications: lisinopril 20mg daily, metformin 1000mg BID, atorvastatin 40mg daily. Denies cocaine use, recent immobilization, or prior VTE.
Assessment & Plan: 1. Acute Chest Pain — High-risk presentation HEART Score: 7 (High Risk) - History: moderately suspicious (2), EKG: non-specific ST changes (1), Age: >65 (2), Risk factors: ≥3 (2), Troponin: pending initial (0) - Risk stratification: >12% 6-week MACE event rate → full ACS workup - Serial troponins q3h, 12-lead EKG (initial + repeat at 60 min) - ASA 325mg PO administered, heparin drip per ACS protocol - Cardiology consulted for probable cath lab activation Dangerous Diagnosis Exclusion: - STEMI: No ST elevation on initial EKG — serial monitoring - Aortic dissection: No tearing quality, no pulse differential, no mediastinal widening on CXR - PE: Low pretest probability (Wells PE: 1.5) — no immobilization, no prior VTE, no hemoptysis 2. Hypertension — BP 168/94 on arrival, improved to 142/88 post-NTG - Continue home lisinopril, recheck prior to disposition 3. DM2 — Glucose 186 on arrival - Hold metformin (contrast exposure risk), sliding scale insulin PRN
Cardiology Consult: 71-year-old female with new-onset atrial fibrillation and decompensated heart failure
HPI: 71 y/o female with known HFrEF (EF 35% on TTE 6 months ago), HTN, DM2, and CKD Stage 3a (baseline Cr 1.4), consulted for new-onset atrial fibrillation with RVR. Progressive exertional dyspnea (NYHA II → IV over 2 weeks), 3-pillow orthopnea (new), PND, and 4+ pitting edema to knees bilaterally. Weight gain of 8 lbs over 2 weeks. Triggers explored: dietary indiscretion (holiday meals), medication non-adherence (ran out of furosemide 10 days ago). No prior AF history, no palpitations previously. Denies chest pain, syncope, or presyncope.
Assessment & Plan: Reason for Consult: New-onset atrial fibrillation with RVR in setting of acute decompensated HFrEF. Requesting Service: Internal Medicine (Dr. Patel) 1. New-onset Atrial Fibrillation with RVR CHA₂DS₂-VASc Score: 5 (Female sex, age ≥75, HTN, DM, HF) - Stroke risk: 6.7%/year → strong anticoagulation indication - Rate control: avoid diltiazem (negative inotropy in HFrEF) → amiodarone 150mg IV bolus then 1mg/min × 6h - Anticoagulation: apixaban 5mg BID (preferred over warfarin given age + fall risk) - TEE prior to any cardioversion attempt if AF duration uncertain - HAS-BLED: 2 (HTN, age) — acceptable bleeding risk 2. Acute Decompensated Heart Failure (ADHF) NYHA Functional Class: IV (dyspnea at rest) - IV furosemide 80mg BID (2.5× home oral dose), net negative 1-1.5L/day - Carvedilol held (acute decompensation), restart when euvolemic - Add sacubitril/valsartan when stable (PARADIGM-HF indication: EF ≤40% on ACEi) - BNP trend: 1,840 → monitor for response to diuresis 3. CKD Stage 3a — Cr 1.4 (baseline), monitor with aggressive diuresis - KDIGO risk: moderate — monitor Cr/K daily with diuresis
Operative Note: Right total knee arthroplasty in 68-year-old male with severe tricompartmental osteoarthritis
Operative Note: Preoperative Diagnosis: Right knee severe tricompartmental osteoarthritis (Kellgren-Lawrence Grade IV) Postoperative Diagnosis: Same Procedure: Right total knee arthroplasty Surgeon: Dr. James Morrison (attending — present and scrubbed for entire case) Assistant: Dr. Chen (PGY-4) Anesthesia: Spinal with sedation (Dr. Rivera) Implants: Smith & Nephew Legion CR, Size 5 femoral / Size 4 tibial / 10mm poly insert EBL: 150 mL Tourniquet Time: 62 minutes (pneumatic, 275 mmHg) Specimens: Femoral and tibial bone cuts — to pathology Complications: None Findings: Severe tricompartmental degenerative changes with exposed subchondral bone medially, grade III chondromalacia laterally, intact PCL
Post-Operative Plan: 1. Weight Bearing: WBAT right LE with front-wheeled walker 2. DVT Prophylaxis: Enoxaparin 40mg SQ daily × 14 days + mechanical (SCDs while inpatient) 3. Pain: Multimodal — scheduled acetaminophen 1g Q6h, meloxicam 15mg daily, tramadol 50mg Q6h PRN (max 14 days) 4. Antibiotics: Ancef 2g IV × 24h post-op 5. PT: Initiate POD0 PM — CPM machine, active/passive ROM, gait training 6. Follow-up: 2 weeks (staple removal + wound check), 6 weeks (X-ray + ROM assessment) 7. VTE Risk: Caprini Score 7 (High Risk) — extended pharmacologic prophylaxis indicated
Primary Care: 55-year-old female Medicare Advantage patient with DM2, HTN, depression, and overdue preventive care
HPI: 55 y/o female Medicare Advantage patient presenting for chronic disease management follow-up. Last visit 3 months ago. Diabetes (DM2, 8 years): A1c 8.2% (up from 7.6%), reports adherence but dietary indiscretion during holidays. Home glucose logs show fasting 140-180 range. No hypoglycemic episodes. Denies polyuria, polydipsia, vision changes, or foot numbness. Last diabetic eye exam: 14 months ago (overdue). Last podiatry visit: never. Hypertension (12 years): Home BP readings averaging 135-145/85-90. Taking lisinopril 20mg daily consistently. Depression (MDD, recurrent): PHQ-9 score today: 14 (moderately severe). Persistent low mood, anhedonia, poor sleep (initial insomnia), decreased concentration. On sertraline 50mg × 6 months with partial response. Denies SI/HI, denies alcohol or substance use. Columbia Suicide Severity: negative for ideation and behavior.
Assessment & Plan: 1. DM2, Uncontrolled (A1c 8.2%) — HCC 19 - Add empagliflozin 10mg daily (SGLT2i — CV and renal benefit, EMPA-REG OUTCOME indication) - Continue metformin 500mg BID (not escalating given GI intolerance history) - Diabetic eye exam referral (overdue 2 months — HEDIS measure) - Podiatry referral for initial foot exam - Recheck A1c in 3 months, target <7% 2. Hypertension, Suboptimally Controlled — HCC (when with CKD/DM) - BP today 138/86 — above target of <130/80 (ACC/AHA for DM patients) - Increase lisinopril to 40mg daily - Home BP log review in 4 weeks 3. Major Depressive Disorder, Recurrent, Moderate — HCC 59 - PHQ-9: 14 (moderately severe) — partial response to sertraline 50mg - Increase sertraline to 100mg daily - Safety plan reviewed, crisis line provided (988) - Follow-up in 4 weeks, recheck PHQ-9 - If inadequate response → consider augmentation or psychiatry referral 4. Preventive Care Gaps Addressed: - Mammogram ordered (last: 26 months ago — HEDIS BCS measure) - Colonoscopy referral (age 55, average risk, never screened — HEDIS COL) - Tobacco screening: former smoker, quit 3 years — MIPS measure 226 - Annual flu vaccine administered today — MIPS measure 110
Record the encounter. Review the draft. Sign.
From bedside to a clinician-ready note without typing through dinner.
Two hours back at the dinner table.
Not a productivity pitch. Not a time-savings promise. A description of what a Tuesday actually looks like once the tool works for you instead of for someone else’s billing.
Tuesday · 11:04 PM
Without Scribeable
- Typing notes at 11pm while your family sleeps.
- Charting 18 patients from memory after rounds end.
- HCC codes the encounter already supported, lost in the narrative.
- A scribe your hospital picked, tuned to their billing targets.
Built by a physician. For physicians.
Scott Kohlhepp, DO
“I’m a physician. I practice internal medicine. I built Scribeable because I know what a great documentation tool needs to do — not from a product roadmap, from doing the job. The tools that existed were designed by product teams who’d never rounded, pitched to hospital IT committees who’d never asked the physicians what they needed.”
“Physicians are getting squeezed from every direction. Insurers deciding what’s medically necessary. Administrators setting documentation requirements. VCs choosing your software. The one thing that should be entirely yours — the clinical note — kept ending up in someone else’s black box. So I built the version I wanted. Two-pass verification. Every sentence cited. 236 calculators re-verified against the transcript. An account that belongs to you, not the hospital.”
“Abridge and DAX are good products — for the hospital IT committees they were built for. I built this for the physician. I have a professional stake in whether it works for you, and I’m here if it doesn’t. Help me build the next piece of it.”
Don’t just use Scribeable.
Most AI scribes ask you to adopt their product. We’re asking you to help decide what it becomes. The Clinician Guild is 20–30 physicians who get a real vote on the roadmap, early access, and public credit when their input ships. No equity exchange, no NDA, no marketing ambassador BS.
A physician’s seat at the table where product decisions get made.
- A real vote on what ships next quarter
- 2–4 weeks early access to every new feature
- Quarterly 60-minute call with the founder
- Named publicly as a Guild member (with consent)
- No equity, no cash, no NDA, no time sink
Composite testimonials are everywhere. We’re tired of them too.
The industry quietly replaces real reviews with “representative experiences” and renamed initials. We pulled ours down. The replacement is coming.
What’s coming
- Every review NPI-verified. Matched to the CMS registry: real name, real license, real specialty.
- Every rating published. 5★ through 1★, honest distribution with counts. No curated positivity.
- Founder responses public. If a reviewer raises a concern, our reply sits next to it, timestamped.
- Pay for reviews. No credits, no discounts, no gift cards, no priority support. Ever. It violates Apple’s §5.6.3 and our own house rules.
- Filter the solicitation pool. We prompt by neutral behavioral criteria (e.g. ≥10 notes saved), never by NPS.
- Hide the bad ones. If someone writes a 1☆ review, it gets published alongside our response.
Want to be one of the first 100 verified reviewers? We’ll reach out to active users directly — no incentive, just honest feedback.
Read the ethics floorHow Scribeable compares
Same sources as every comparison page on this site. One table, one source of truth.
| Capability | Scribeable | Nuance DAX | Abridge | Suki AI |
|---|---|---|---|---|
| Clinical Calculators | 236 with smart activation + backend validation | None | None | None |
| Quality Measures | 56 (MIPS, HEDIS, CMS eCQMs) | None | None | None |
| HCC Crosswalk | 8,400+ entries, built in | Add-on | No | Limited |
| MIPS Scoring | Real-time with payment adjustments | No | No | No |
| Two-Pass Verification | Stage 1 draft → Stage 2 verifies + corrects | Single pass | Single pass | Single pass |
| Continuity of Care | Pre-visit briefs, cross-encounter context | No | No | No |
| Note Generation Speed | 60 seconds | 3–5 minutes | 2–4 minutes | 2–3 minutes |
| Starting Price | Free ($0) | Enterprise only | Enterprise only | $199+/mo |
| Specialties | 46 | Limited | Limited | Limited |
| After Visit Summaries | Yes (46 specialties) | No | No | No |
| Apple Watch | Yes (native) | No | No | No |
| HIPAA + BAA | All paid plans, no extra cost | Enterprise only | Enterprise only | All plans |
| Setup Time | 5-minute browser extension | 6–12 month IT integration | 6–12 month IT integration | Weeks |
Capability statements are lifted from public marketing materials and reconciled against Scribeable's platform state as of April 2026. Verify with your implementation lead before procurement.
Radical Transparency
Read the architecture. Audit every BAA. Leave whenever you want.
HIPAA is the floor. These are the five layers on top of it. Every BAA we’ve executed is listed at /transparency, including the upstream LLM. Your audio stays on-device. Your data never trains a model, ours or anyone else’s.
API Request Encryption
Your notes and patient information are encrypted before they ever leave your device.
API Response Encryption
Generated notes are encrypted on the server before being sent back to you.
WebSocket Message Encryption
Live transcription during patient encounters is encrypted in real time.
Audio Stream Encryption
Encounter recordings are encrypted before leaving your device.
Field-Level Storage Encryption
Each data field is independently encrypted at rest with per-organization keys via Google Cloud KMS.
Revenue
You already did the work. We help you bill for it.
Most missed revenue isn’t missed care — it’s missed documentation. HCC codes the chart supports but the note never named. E&M levels under-coded because the MDM wasn’t written out. Quality measures the encounter met but the billing line didn’t carry.
A single missed HCC code can represent $800–$3,000 in annual risk-adjusted revenue per Medicare Advantage patient.
CMS, 2024 HCC risk-adjustment factor payment data. The number you actually recover depends on your panel, your payer mix, and how often your current documentation names the condition the chart already supports. Scribeable’s HCC V28 engine flags codes your encounter already supports — so you can choose to add them. The physician decides what goes in the note. We just stop letting revenue slip out of it.
Your notes shouldn’t train the next scribe
Your workflow shouldn’t be chosen by someone who has never touched a patient. Your billing codes shouldn’t be decided by whichever vendor bundled with your hospital’s EHR contract. Your documentation shouldn’t belong to Silicon Valley the day you sign a BAA with it.
So this is what we’re building — a clinical documentation platform that earns its place in your workflow at every price point (solo, practice, and enterprise), support from people who actually practice medicine, and a seat for clinicians at the table where the tool gets shaped.
“We’re not building an AI scribe.
We’re building the operating system for what’s left of private medicine.”
Fair pricing for how physicians actually work.
Solo, practice, or enterprise — same engine, same architecture, priced for the work. 14-day trial, no credit card, no sales call.
Still not sure? Start with a 14-day free trial. No credit card. No sales call. No commitment. And if you want to help us build the next piece of this, the Guild takes 30 physicians a year.