Optometry revenue cycle, stage by stage
Automating an optometry revenue cycle with RPA
If you run an optometric practice like The Eye Center Doctors of Optometry, P.A. and you are weighing RPA for the revenue cycle, the useful question is not whether to automate. It is which stage actually pays, and which stage has no API behind the screen.
Direct answer · verified June 21, 2026
The optometry revenue cycle gets automated one desktop stage at a time, and the stage that pays most is cash application.
In the one published optometry RPA story, MyEyeDr calls cash application their largest robotic-process-automation use case. It touched $186 million in remittances in a single year. Their software got them about 30% of the way; RPA picked up another 40% to 50%. That remaining half lives on practice-management screens with no write-back API, which is exactly where the choice of automation engine decides whether it survives the next interface update.
What this question is really asking
A practice the size of The Eye Center Doctors of Optometry, P.A. runs the same revenue cycle a thousand other multi-doctor optometry groups run: verify benefits, code the visit, send the claim, post the remittance, work the denials, chase the patient balance. The P.A. in the name is the Professional Association entity type, not the state of Pennsylvania, which trips up a lot of searches for practices named this way.
When someone pairs a practice like this with “automation, RPA, revenue cycle” they are usually doing one of two things: benchmarking how a group of this size automates back-office billing, or deciding whether to buy RPA for their own cycle. Both questions have the same answer, and it is not a feature list. It is a map of where the cycle stalls.
The six stages, and where each one stalls
Walk the cycle in order. Every stage can be automated. What changes stage to stage is how much of the work sits behind an API versus how much falls to a person typing into a screen.
Eligibility and benefits verification
Front desk pulls vision and medical benefits before the visit. Vision benefits live in a separate plan portal from medical, so one patient can mean two logins and two screens to read.
Where it stalls: The plan portals rarely expose a benefits API to a small practice. A bot has to read the portal screen the same way a human does.
Charge entry and coding
Exam, materials, and any medical procedure get coded and entered as charges in the practice-management system after the encounter closes.
Where it stalls: Charges are typed into the PMS chart. There is usually no write-back endpoint to push a coded charge into the ledger programmatically.
Claim submission
Claims go out to vision plans and medical payers, often through a clearinghouse, sometimes keyed straight into a payer portal for the plans the clearinghouse does not cover.
Where it stalls: The clearinghouse handles its own roster of payers. The leftover payers get keyed by hand into portals that have no submission API.
Cash application and ERA posting
Remittances come back, payments and adjustments get posted line by line against the right claim. This is the highest-volume, highest-value stage of the whole cycle.
Where it stalls: This is where the API gap hurts most. The remittance has to be posted into the PMS screen. MyEyeDr calls cash application its single largest RPA use case.
Denial management
Denied and underpaid claims get worked: read the denial reason, fix the code or the eligibility detail, resubmit. Optometry leaks a large share of revenue here.
Where it stalls: Working a denial means jumping between the payer portal, the clearinghouse, and the PMS, none of which share state through an API.
Patient AR and statements
Patient responsibility gets billed, statements go out, balances get followed up. The long tail of the cycle that quietly ages into write-offs.
Where it stalls: Statement generation and posting of patient payments land back in the same no-API PMS screens as everything else.
Stage 04 is the one that pays: cash application
The single most useful number in optometry revenue-cycle automation comes straight from the MyEyeDr customer story. Cassie Haag, their Director of Systems and Analytics, describes cash application as their largest RPA use case and says it touched $186 million in remittances in one year. Then she puts a number on the integration gap most RCM pitches skip: the software gets them roughly 30% of the way, and RPA picks up another 40% to 50%, with staff doing the rest.
“Cash application is our largest RPA use case. It's touched $186 million worth of remittances this year. Our software gets us 30% there, and RPA picks up where the software leaves off and does another 40% to 50%, with staff doing the rest.”
Cassie Haag, Director of Systems and Analytics, MyEyeDr (UiPath customer story)
Read that 40-to-50% from a buyer’s seat. Half the cash-application work lives in a layer the practice-management vendor will not hand you an API for. The remittance still has to be posted line by line into the ledger screen. So whichever engine you pick, the real job at this stage is the same: drive the posting screen reliably, every day, without a developer babysitting it.
Figures reported in the MyEyeDr UiPath customer story. MyEyeDr grew from 40 vision centers to over 850, largely by acquisition, which is what put this volume of remittances and claims through automation.
Why the API gap shows up only on the return trip
Trace one claim through the cycle. On the way out, a clearinghouse can do a lot for you. On the way back, the remittance has to be posted into a practice-management screen that almost never accepts a programmatic write. That return trip is the no-API leg, and it is the leg with the most volume.
One claim, out and back
The red leg is the one that has to be typed. It is also the highest-volume step in the cycle, which is why cash application is where automation earns its keep.
What a desktop agent actually sits between
A revenue-cycle automation on the desktop is not a single integration. It reads from the places benefits and payments live, and it writes back into the practice-management system through the same screens a biller uses. None of these handoffs go through a clean API, which is why a screen-reading approach is the one that reaches all of them.
Revenue-cycle desktop agent: reads, then posts
The thing that decides whether a posting bot survives an update
Once you accept that the posting screen has to be driven, the only question left is how the automation finds a field. Selector-driven and pixel-driven RPA record where a field sat during the build. When the PMS ships a layout change, that location is wrong and the bot stops until someone re-records it. For a cycle posting thousands of lines a day, every stall is real money and aging AR.
Mediar reads the accessibility tree the operating system already exposes, the same interface a screen reader uses, and resolves a field by its label. A field named “Payment Amount” keeps resolving even if it moves. The open-source Terminator SDK (github.com/mediar-ai/terminator) is the layer that does this, and the shape of it is small:
The point is the absence of a coordinate or a recorded selector anywhere in that snippet. The field is found by the name the screen already advertises, which is what makes the posting automation self-heal when the interface shifts.
When you do not need any of this
If your RCM vendor already posts cash cleanly through a direct integration with your specific PMS, and your denials are low, leave it alone. Automation earns its place where volume is high and the API is missing, not everywhere. The MyEyeDr numbers are real and they were earned on UiPath, so an existing UiPath Center of Excellence with working posting robots is not something to rip out on principle.
The case to look at Mediar is narrower: your posting and denial work is the bottleneck, your robots keep breaking on PMS layout changes, or you are staring at a six-figure implementation quote to automate a handful of screens. That is the 40-to-50% the optometry case study itself points at, and it is the layer the accessibility-tree approach was built for. At $0.75 per minute of runtime with no per-seat licensing, the math is usually about how much posting volume you have, not how many bots you license.
See your ERA posting screen automated by element name
Bring one remittance batch and your PMS. We will show it posted line by line, found by field name rather than selectors, on a short call.
Optometry revenue cycle automation: common questions
How is the optometry revenue cycle automated with RPA?
Stage by stage on the practice-management desktop: eligibility checks, charge entry, claim submission, cash application and ERA posting, denial rework, and patient AR. The biggest single win is cash application. In the MyEyeDr UiPath case study it is described as their largest RPA use case, touching $186 million in remittances in one year, with the software getting them about 30% of the way and RPA picking up another 40% to 50%.
Which stage of the optometry revenue cycle is hardest to automate, and why?
Cash application and ERA posting. It is the highest volume, and it is also the stage most exposed to the missing API. The remittance has to land back inside the practice-management ledger, and optometry PMS platforms rarely give you a write-back endpoint for that screen. So the only way in is to drive the screen the way a biller does, which is why screen-driven automation matters most exactly here.
Does 'P.A.' in 'The Eye Center Doctors of Optometry, P.A.' mean Pennsylvania?
No. In a practice name like this, P.A. is the Professional Association legal structure that US optometry and medical groups commonly incorporate under. It describes the business entity, not the state. Several optometry groups across different states carry the P.A. suffix.
Why can't a clearinghouse or RCM vendor just integrate by API and skip the screen work?
Clearinghouses cover their own roster of payers, and good RCM software handles a real share of the routine work. The MyEyeDr team puts a number on the gap: their software gets them roughly 30% of the way. The remaining 40% to 50% lives in screens and portals with no write-back API, which is the work that falls to a person typing, and therefore the work that screen-driven automation has to cover.
What happens to an optometry posting bot when the PMS ships a UI update?
With selector-driven or pixel-driven RPA, the bot misses and stops until a developer re-records the step. Mediar reads the accessibility tree the operating system already exposes and resolves a field by its label, so a field named 'Payment Amount' keeps resolving even if it moves. There are no brittle selectors to maintain, which is the difference between a posting automation that survives the next update and one that does not.
Is screen-driven automation safe for patient and payment data in an eye-care practice?
Mediar is SOC 2 Type II certified and HIPAA compliant, and deploys on-prem or in the cloud. That matters for a revenue cycle handling patient demographics, insurance detail, and payment data. It also keeps audit logs and validation rules on what the automation reads and writes.
Related reading
The Eye Center Doctors of Optometry, P.A. UiPath case study
The honest answer on whether a UiPath case study exists for this practice, and the optometry automation story that actually does.
EMR billing workflow automation
Where billing automation breaks on the EMR desktop and how field-by-name automation survives interface drift.
Healthcare automation software
Why no-API legacy systems are exactly where browser-based AI agents do not help, and accessibility-tree automation does.
Comments (••)
Leave a comment to see what others are saying.Public and anonymous. No signup.