The desktop app · Assistants
Assistants that read the chart.
An assistant is a page of plain-English instructions the AI follows for every chart — what to include, what to skip, how the note should read. It runs on your own machine against de-identified records, so no AI ever sees a name.
Below: each assistant, shown by what it produces. Hover an output to see exactly what the AI saw. Synthetic patient data.
Before the visit
Chart prep waiting when the day starts.
OUTPUT · FOLLOW-UP · 8:00 AM⛨
LAST 3 MD VISITS — 07/10, 04/02, 01/15
T2DM — metformin 1000mg BID; A1c 8.2%, up from 7.8%
HTN — lisinopril 20mg daily; BP 148/92 in office
Open: A1c recheck promised 04/02 — not ordered
Follow Up Visit Summary
Customer createdLast three MD visits and a problem-based summary, as consistent plain text ready to paste into the HPI.
OUTPUT · ACUTE · 2:15 PM⛨
Reason: cough × 5 days, worse at night
Relevant: asthma — ICS daily; last flare 11/2025
Allergies: penicillin (hives) · Meds: albuterol PRN
Recent: CXR 01/2026 clear
Sick Visit Prep
Customer createdReason and symptoms, relevant active problems, recent findings, current meds and allergies — before an acute visit.
OUTPUT · NEW PATIENT · THU⛨
T2DM (2019) — metformin 1000mg BID · A1c 7.4 (11/2025)
Hypothyroid (2014) — levo 75mcg · TSH 2.1 (09/2025)
Gaps: colonoscopy never done (age 54) · lipids 14 mo overdue
New Patient History
Customer createdActive conditions grouped with meds and monitoring, plus health-maintenance gaps — every item source-dated.
Documentation
Drafts grounded in documented findings only — you own the final word.
OUTPUT · WOUND CLINIC · RM 3⛨
NARRATIVE — L lateral malleolus ulcer 2.1 × 1.4 cm,
granulating; down from 2.8 × 1.9 cm on 07/30.
ASSESSMENT — Improving venous ulcer; no infection
signs documented
Wound Care Visit Note
Customer createdOne-pass Narrative, Assessment and Plan from the record, tracking progression across prior visits.
OUTPUT · REFERRAL · CARDIOLOGY⛨
QUESTION — exertional dyspnea despite optimized BP
Workup: TTE 06/14 (EF 55%), troponin neg ×2, CXR clear
Active meds: lisinopril 20mg, metoprolol 25mg BID
Specialist Referral Summary
Relevant history, completed workup, active medications, and the specific clinical question being asked.
Back office
Prior auths, appeals and outreach assembled from the chart — staff review and send.
OUTPUT · PA PACKET⛨
ORDER — MRI lumbar spine
Failed: naproxen 6 wks (03/2026), meloxicam 6 wks (05/2026)
PT completed — 8 sessions, symptoms persist
Prior Authorization Packet Assembly
The clinical justification for a pending order — notes, labs and failed prior therapies supporting medical necessity.
OUTPUT · APPEAL · CLAIM #4417⛨
RE — denial CO-50 (not medically necessary)
Supporting: office note 04/18, failed conservative tx;
imaging report 05/02. Letter drafted for review.
Claim Denial Appeal Drafter
Reads a denied claim, finds the supporting documentation, and drafts a formal appeal letter grounded in it.
OUTPUT · FRONT DESK · TOMORROW⛨
1. Reyes, M — A1c overdue 6 mo; promised 04/02
2. Okafor, J — cardiology referral unbooked 8 wks
3. Tran, L — annual wellness unscheduled
Front Desk Outreach Call List
Unresolved care gaps — undrawn labs, unbooked referrals, unscheduled follow-ups — as a prioritized call list.
OUTPUT · OUTREACH QUEUE⛨
1. Colorectal screening — never done, age 54
“We’d like to get your screening on the books…”
2. Flu vaccine — not documented this season
Quality Gap Outreach Prioritizer
Open quality gaps ranked by clinical urgency, with a one-line caller script for each overdue item.
The assistant builder
Write your own.
Describe the output you want in a paragraph. The builder sets up the data, prompt, output and guardrails itself — what you wrote is what it does, and you can read it any time.
Before each physical, list which screenings are due by age and history, with the last completed date for each. One line per item, most overdue first.
✓ Data, prompt, output and guardrails — set up from this paragraph