PREPARED FOR ·
Faxes and referrals get read and filed. Prior-auth packets get assembled. Patient messages get sorted. A report lands on the front desk each morning. It runs in your own servers or your own AWS account.
The work is rarely the problem. The hunting, retyping, and chasing around it is.
Every request needs forms, notes, and follow-up calls. Staff re-key the same details into payer portals.
Someone opens each one, finds the patient, and files it by hand. A referral that lands on the wrong chart is easy to miss.
Portal, phone, text, and email. The urgent one is buried under refill requests and billing questions.
Paperwork and rework cost every practice time that could go to patients.
Source · CAQH Index (via Dataspring), 2025
HIPAA applies, and a business associate agreement is required. Any AI that touches patient information has to run under one, in an environment you control.
A day in the life. Pick a use case and watch the question or report come together, with citations back to the source.
Each fax is read with OCR, matched to the right chart, and filed. Anything uncertain goes to a person.
Sample patients are shown by initials. In your office, access follows your EHR roles.
| Document | Matched to | Action |
|---|---|---|
| Cardiology referral [1] | Chart J.M. | Filed to Referrals |
| Lab result, panel 2 [2] | Chart A.T. | Filed. Flagged for provider. |
| Page 2 only, no header [3] | No match | Held for front desk |
The assistant gathers the order, visit notes, referral, and insurance details into one packet. It shows what is missing before the payer does.
It assembles and checks. The provider decides what goes to the payer.
Messages from the portal, texts, and transcribed voicemails are sorted by topic and urgency. A nurse sees the urgent queue first.
Suggestions only. A nurse reviews the urgent queue. No message is answered automatically.
| Message | Channel | Suggested queue |
|---|---|---|
| Chest tightness since last night [1] | Portal | Nurse, now |
| Refill request, blood pressure [2] | Text | Refill queue |
| Can I move Thursday's visit? [3] | Voicemail (transcribed) | Front desk |
| Question about a bill [4] | Billing |
A one-page report before the doors open: what came in, what is waiting, and who needs a call.
Runs on your schedule. People decide what to act on.
Not just documents. Scans and photos are read with OCR. Phone calls are transcribed. Texts are included when you choose.
Most firms don't know AI can do this. Patient lists collect duplicates and typos for years. The assistant finds likely matches, fills gaps from other records, and flags what a person must confirm.
| Name | Phone | Insurance ID |
|---|---|---|
| WEBB, MARCUS | 7045550188 | BCBS-4471-2290 |
| Marcus L Webb | 704-555-0188 | BCBS44712290 |
| Priya Nair | (980)555-O112 | (blank) |
| nair, priya | 980 555 0112 | Aetna W2209 |
| T. Okonkwo | 7045550165 | Self-pay |
| Lena Park | 7O4 555 0177 | UHC 88-1204 |
Messy rows: mixed formats, typos, duplicates, and blanks. Select After, or Replay, to watch it clean up.
It is a help to your whole team, not a replacement. Everyone on your team gets an assistant that knows the context of your work. People still make the calls.
Answers follow the roles already set in your EHR and file shares. Built around minimum-necessary access.
A provider can give a covering nurse or float staff access for a set time. You decide who, and for how long.
Front desk, nurses, providers, and billing see different slices of the same index.
A record of who asked what and which items were opened, to support your access reviews.
Nothing is hidden for this role.
| Time | Who | What |
|---|---|---|
| 07:52 | R. Dalton, front desk | Opened: referral status, J.M. |
| 08:10 | Nurse line | Asked: urgent messages since 5 PM. 4 sources. |
| 09:03 | Admin | Granted covering-nurse access until Friday |
Every question and every opened document is recorded. You can review it or export it.
You choose. Either way, it runs inside an environment you own. Pick one to see the diagram.
Healthcare engagements require a signed BAA. No work touches protected health information until it is in place. On AWS, you sign your own BAA with AWS.
Systems are designed to support your duties under HIPAA. Your counsel and compliance staff decide what your rules require.
From first call to a system your team uses every day. Four steps, and you can stop after any of them.
Free 30 minutes on your EHR, fax flow, payer portals, and inboxes.
We review those systems and how clean the data is. You get a written plan with phases, priorities, and a timeline. A signed BAA comes before any work touches PHI.
We build what the plan lays out, such as fax filing or prior-auth packets, one phase at a time.
Optional ongoing support for new connectors, updates, and help while you care for patients.
You pay your own cloud and model costs directly. We never mark them up or hold your infrastructure.
Book a 20-minute walkthrough. I'll show these pieces on your kind of data, and we'll talk about where the time goes.
I'm a solo consultant with hands-on experience leading enterprise AI search and assistant rollouts. You work with me directly.