For most medical practices in 2026, the first things worth automating are prior authorization prep, billing follow-up, patient intake and appointment reminders. These are the jobs where staff spend hours copying data between the EHR, payer portals and phones. A single workflow costs from $2,500 and usually goes live in 2 to 6 weeks; a prior authorization build is a larger job, with a sample range of $6,000 to $9,000.
This guide is for independent and small group practices: internal medicine, family medicine, specialty groups and their billing teams. If you run a dental office, our dental practice automation guide is the better read.
What changed in 2026: the CMS prior authorization rule
The CMS Interoperability and Prior Authorization Final Rule (CMS-0057-F) applies to Medicare Advantage organizations, state Medicaid and CHIP fee-for-service programs, and Medicaid and CHIP managed care plans. It does not apply to qualified health plans on the federal exchanges in the same way. Per the CMS fact sheet, read October 29, 2026:
- Since January 1, 2026, affected payers must decide expedited prior authorization requests within 72 hours and standard requests within 7 calendar days.
- Payers must give a specific reason when they deny a request.
- Payers must publish prior authorization metrics each year, starting in 2026.
- Prior authorization APIs are required from January 1, 2027.
- Drugs are excluded from these rules.
What it means for a practice: decisions come back faster from these payers, so the slow part becomes your own prep. Gathering notes, codes and attachments for each request is still manual in most offices. That is the part automation can take over. Denials now come with a specific reason, which makes it easier to route them to the right person with the right fix.
Five workflows to automate first
| Workflow | What it replaces | Typical systems |
|---|---|---|
| Prior authorization prep | Staff collect notes, codes and payer forms by hand | EHR, payer portals, FHIR APIs |
| Billing and AR follow-up | Calls and letters for unpaid balances | Practice management, SMS, email |
| Denial triage | Reading each denial and deciding what to do | Clearinghouse, EHR |
| Digital intake | Paper forms retyped into the EHR | Online forms, EHR |
| Reminders and confirmations | Front desk calls the day before | Scheduling, SMS |
1. Prior authorization prep
An automation pulls what the payer needs from the chart, fills the request and flags anything missing for a person to check before it is submitted. Staff still review and submit; the automation does the gathering.
In our prior authorization case study, a 4-physician internal medicine group cut time per request from 12 minutes to 2, and daily authorization time from 3 hours to 40 minutes. The practice estimated it avoided a hire worth about $60,000 a year. Our post on prior auth automation covers the build.
2. Billing and AR follow-up
Statements, reminders and payment links go out on a schedule, and replies route to billing staff. In our billing case study, an 8-provider multi-specialty clinic went from 68 to 41 days in AR over 6 months, and balances older than 90 days fell 54%. See billing follow-up automation.
Tell us the task and your EHR. We'll reply with one idea worth building first.
3. Denial triage
An AI agent reads each denial, classifies the reason and drafts the next step, such as a corrected claim or an appeal letter, for a biller to approve. In our claim denial case study, a 5-person billing team raised its appealable denial recovery rate from 40% to 71% and handled 3 times the claim volume with the same headcount.
4. Digital intake
Patients fill forms on their phone before the visit, and the answers land in the EHR without retyping. In our digital intake case study, new patient processing dropped from 20 minutes to 6 at a 3-location group. That client was a dental organization, but the same pattern applies to medical intake. More in patient intake going digital.
5. Reminders and confirmations
Texts confirm appointments and offer a reschedule link. HHS guidance allows practices to communicate with patients electronically with reasonable safeguards, such as limiting the details in a message. See HHS FAQ 570 and our post on HIPAA-safe appointment reminders.
Getting HIPAA right
Automation does not change who is responsible for patient data. Three rules we follow:
- Every vendor that touches patient data needs a business associate agreement with you. We list each one so you can check.
- Where possible, the automation passes record IDs instead of names or notes, and AI prompts get only what the task needs.
- Risky actions, such as sending clinical details or submitting a claim, wait for a person to approve them.
Our security page has the full rules, and the glossary entry on HIPAA compliance automation explains the terms. Our post on HIPAA and AI tools covers which AI services offer BAAs.
Connecting to your EHR
Most EHRs offer some mix of an API, FHIR endpoints, exports and HL7 feeds. Which one we use depends on your EHR and what your vendor allows. Our EHR integration options post compares them, and FHIR is explained in the glossary.
What it costs
| Build | Price | Time |
|---|---|---|
| Single workflow, such as reminders or billing follow-up | From $2,500 | 2 to 6 weeks |
| Prior authorization prep | $6,000 to $9,000 (sample range) | 4 to 6 weeks |
| First AI agent, such as denial triage | $4,000 | Depends on scope |
| Care plan after launch | $350 a month | Month to month |
The cost guide explains what moves a quote, and every price is on the pricing page. AI model usage is billed by the provider directly, about $30 to $150 a month for most small businesses.
A 90-day plan
- Month 1: pick one workflow with a clear before number, such as minutes per prior auth or days in AR. Write down the current figure.
- Month 2: build and test it on real past cases, with a person reviewing every output.
- Month 3: measure against the before number, then relax review only where the error rate is low. Our AI agent governance checklist covers when that is safe.
The healthcare automation checklist is a printable version of what to check first.
How this fits implementation work
We build these as workflow automation when the steps are fixed and as AI agents when the input varies, such as denials. Our healthcare page lists all healthcare work and case studies. To see what to automate first in your practice, describe one task through the free automation audit. We reply within one business day with one specific automation idea.
Independence and limitations
This guide summarizes public CMS and HHS sources as of October 29, 2026. It is not legal or compliance advice; check with your compliance adviser. Case study results come from specific clients and will differ for your practice.
