The Challenge
Texas Sinus Center is a multi-brand ENT, allergy, and sleep medicine practice serving the greater Houston area. Their surgical operations span Memorial Hermann The Woodlands, Townsen Humble, Legend Hospital, and Woodlands Specialty Hospital. A team of 6+ billing specialists manages the full revenue cycle — from insurance verification to claims submission to denial appeals.
The scale of the operation was significant: 1,410+ bills in progress at any given time, 166 claims in dispute (23% of the total portfolio), and a backlog of 72+ denied claims awaiting appeal. But it wasn't the volume that was the problem — it was how the work got done.
Manual Portal Navigation Across 15+ Payer Systems
Every insurance verification required logging into a different portal — UnitedHealthcare, BCBS Texas, Cigna, Aetna, Carelon, eviCore, TRICARE, Availity, and a dozen more. Staff would copy a policy number from ModMed EMA, paste it into the payer portal, navigate through benefit details, then transcribe everything back into a ModMed sticky note by hand. Each verification took 30-45 minutes.
Authorization by Fax and Manual Transcription
Prior authorization approvals arrived via eFax as PDF documents. Staff would open each fax, read the authorization number, CPT codes, and validity dates, then manually type everything into ModMed's authorization fields and upload the PDF as an attachment. A purely manual, error-prone process repeated dozens of times per day.
Zero Payment Integration
Credit card payments processed through an external terminal didn't sync to ModMed at all. Billing staff reconciled from PDF batch reports, manually posting each payment. Miss a PDF, miss a payment.
4-Source Appeal Assembly
Preparing a single denial appeal — especially for BCBS Texas Claim Review Forms — required pulling data from four separate systems simultaneously: Availity appeal history, ModMed clinical notes, the BCBS portal, and payer remittance PDFs. Each appeal took 40+ minutes to assemble.
Staff estimated losing 4+ minutes per hour just waiting on ModMed's loading screens. Remote workers faced additional connectivity delays. And the institutional knowledge that held it all together? Stored in ModMed sticky notes and macOS desktop stickies — fragile, unsearchable, and impossible to scale.
The Solution
BAM AI didn't start with assumptions. We started with evidence.
Over the course of a full operating week (February 18–25, 2026), BAM AI captured and analyzed 383 screen recordings spanning 17+ hours of real billing work across three workstations and all six billing team members. Every click. Every portal login. Every manual copy-paste. Every 4-minute loading screen.
The result was a complete operational map of the practice's revenue cycle — the first of its kind built entirely from observed behavior rather than interviews or surveys.
From that analysis, BAM AI designed and began deploying six purpose-built AI agents:
1Insurance Verification Agent — Connects to payer databases and auto-populates ModMed with benefits, copays, deductibles, auth requirements, and estimated patient responsibility. Verification drops from 30-45 minutes to under 2 minutes.
2eFax Authorization Agent — OCRs incoming authorization fax PDFs, extracts auth numbers, CPT codes, and validity dates, and writes directly to ModMed authorization fields with the source document attached. No manual transcription.
3Multi-Portal Claim Status Agent — Monitors Availity, UHC, Cigna, BCBS, and other portals simultaneously. Updates ModMed billing notes automatically and surfaces only the claims that need human attention.
4ERA Reconciliation Agent — Posts electronic remittance advices with automatic fee schedule comparison, secondary claim detection, and discrepancy flagging — replacing the ModMed-Preview-Calculator triangle.
5Surgical Schedule Scrubbing Agent — Pre-checks every surgical patient's eligibility, authorization, and facility credentialing status 2 weeks ahead. Gaps are flagged before they become day-of-surgery cancellations.
6Denial Appeal Assembly Agent — Pulls data from all four required sources, pre-drafts BCBS Claim Review Forms and medical necessity letters, and presents a complete appeal package for 5-minute human review.
The Results (Projected — 90-Day Deployment)
| Metric | Before | Projected | Impact |
|---|---|---|---|
| Insurance verification | 30-45 min/patient | < 2 min/patient | 93% reduction |
| ERA posting throughput | 14 patients/hr | 50+ patients/hr | 3.5× faster |
| Claims in dispute | 23% of portfolio | < 8% | 65% reduction |
| Denied claims backlog | 72+ claims | < 20 claims | 72% reduction |
| Auth transcription | 10-15 min/auth | < 1 min/auth | 90% reduction |
| Appeal preparation | 40+ min/appeal | 5 min review | 87% reduction |
| Annual revenue recovered | — | $350K–$500K | Faster collections + fewer denials |
| Staff hours saved | — | 80+ hrs/week | Redeployed to patient care |
"BAM AI didn't just watch our workflows — they recorded every click across our entire billing department, found inefficiencies we'd been living with for years, and built AI agents that actually understand how medical billing works. The level of detail in their operational analysis was unlike anything we've seen from any vendor."
— Testimonial pending client approvalFrequently Asked Questions
How does BAM AI automate insurance verification for medical practices?
BAM AI's Insurance Verification Agent connects directly to payer databases (UnitedHealthcare, BCBS, Cigna, Aetna, and 15+ others), auto-populates your EHR with benefits, copays, deductibles, authorization requirements, and estimated patient responsibility. Verification drops from 30–45 minutes per patient to under 2 minutes — a 93% reduction. Learn more about AI agents for medical practices.
How much revenue can AI billing agents recover for an ENT practice?
Based on this deployment managing 1,410+ active bills, projected annual revenue recovery is $350K–$500K through faster claims processing, 65% fewer disputed claims, and automated denial appeals. See how AI agents work for ENT practices and healthcare organizations.
What is BAM AI's screen recording analysis method?
BAM AI captures real billing workflows — every click, portal login, and delay — then builds purpose-built AI agents targeting the exact bottlenecks observed. For this practice, 383 recordings spanning 17+ hours were analyzed. See how this approach scales for hospitals and dermatology clinics.
Related Solutions
- AI Agents for Medical Practices
- AI Agents for ENT Practices
- AI Agents for Hospitals
- Healthcare AI Automation
- Frequently Asked Questions
This case study is based on BAM AI's operational analysis of Texas Sinus Center's billing operations conducted February 18–25, 2026. Projected results are based on observed workflow timings, industry benchmarks, and BAM AI deployment data. Actual results will be reported after the 90-day deployment period.