Every optometry patient is two coverage checks and two billing tracks. Relay's AI employees run both, across every location.

A failed eligibility check on the date of service is the single most preventable denial source in optometry. A lapsed plan, an already-used benefit, or a misidentified Medicare Advantage carve-out produces a claim that will be denied and require rework after the patient has gone home.

Relay's AI employee runs a nightly eligibility sweep across both tracks for everyone scheduled in the next 48 to 72 hours, flags coverage gaps, inactive benefits, and already-used vision allowances, and surfaces a prioritized worklist each morning. A staff member resolves the exceptions before the patient arrives, not after the claim is denied.

Optometry verification is two jobs, not one. Routine vision exams bill to a vision plan (VSP, EyeMed, Davis Vision, Spectera, Superior Vision, Humana Vision); medically necessary visits bill to medical insurance. Both must be verified before service. Vision plans run their own portals that most practice management systems do not integrate natively, and Medicare Advantage vision benefits are often carved out to a third-party manager, so staff must know which plan handles which service before the patient sits down.

Coverage gaps get caught before the patient arrives, not after the claim is denied.

Manual prior-auth tracking across several sites against a shrinking decision window is a time-bound revenue risk. An expired auth on a recurring medical-retina protocol is a recurring, avoidable write-off.

Relay's AI employee scans the upcoming schedule against payer-specific auth rules, flags which visits need authorization, drafts the submission packet (clinical notes, ICD-10, CPT, ordering provider), and routes it for one-click submission by an authorized staff member. It monitors open requests against the decision window, alerts staff when a pending auth approaches the visit date, and writes the auth number back to the patient record.

Medical eye visits increasingly require prior authorization: glaucoma, diabetic eye disease, macular degeneration, low vision, and newer diagnostics like OCT and fundus photography. Requirements vary by payer, plan, and CPT. Multi-location groups multiply the problem: an approval logged at the main office may never reach a satellite's workflow, so the satellite bills without the auth number and gets denied.

CMS-0057-F decision-timeframe requirements take effect January 1, 2026: standard prior-auth decisions within 7 calendar days and expedited decisions within 72 hours. Watching the decision window across every location removes the manual calendar-tracking that fails first when one site gets busy.

Open auths get watched against the decision window across every location, not just the main office.

Coding confusion between the 920xx vision track and the 992xx medical track is the documented leading source of optometry claim denials. A claim sent to the wrong payer can expire past its timely-filing window before anyone catches it.

Relay's AI employee runs a pre-submission scrub on every batch: validates that the service type maps to the correct payer, checks the 920xx-vs-992xx split, reviews modifier 25 and 59 use and ICD-10 laterality, confirms the auth number is populated where required, and flags any claim approaching its timely-filing window. A staff member reviews the flagged list and releases the batch. The AI gates the queue; it does not submit autonomously.

Add modifier 25 and 59 use, ICD-10 laterality specificity, and contact-lens fitting-fee-vs-material claims, and the ways a claim can fail multiply. High-volume central billing teams running several locations do not catch every one in time. This is downstream pre-billing QA that begins when the note is signed, not exam-room scribing.

Note on scope: this pre-billing QA reads the signed note for coding accuracy; it does not author the note. Relay handles the downstream claim, while AI scribes handle the exam room.

The coding split gets caught before the claim goes out, not after it denies.

Vision-plan filing windows are often shorter than medical windows, so a denial that sits unworked can expire into a permanent write-off. Multi-location volume amplifies every error rate.

Relay's AI employee triages the denial queue daily across both tracks, categorizes by root cause (eligibility, coding, modifier, documentation, auth), drafts the correction or appeal for common patterns, and routes a prioritized action queue to the billing coordinator who reviews and submits. Denial patterns are surfaced to the operator monthly by payer, code, and location, so the worst-performing site stops being invisible.

Optometry denial patterns are predictable but labor-intensive. Vision-plan denials cluster around benefit-used, non-covered, out-of-network, and missing Rx. Medical denials cluster around missing medical necessity, missing or expired auth, wrong place of service, and coordination-of-benefits errors. Working each one means looking up the reason, pulling the original claim and documentation, correcting, and resubmitting by hand, against payer rules that change without notice.

Denials get worked before the appeal window closes, not after.

The biggest visibility gap for multi-location optometry groups is not the individual denied claim; it is not knowing which site is accumulating AR and why. RevolutionEHR's ERA handling requires manual intervention when submissions hiccup; Compulink's ledger system struggles to run cleanly across three or more locations; Eyefinity's financial reporting does not surface totals comparably across sites. Each platform is doing its job, but none of them was designed to show you the full picture across all your locations.

Relay's AI employees consolidate ERA posting reconciliation, flag unmatched remittance, identify aging buckets by payer and location, and surface the action list: what needs a correction, what needs a call, what needs a resubmission. A staff member reviews and works the queue. You get one view of your AR across all locations, without replacing the EHR.

This is not a software limitation; it is a workflow gap that grows with every location you add.

AR across every location becomes visible in one place, without replacing the EHR.

The patients who never rebooked and the vision benefits about to reset are invisible without someone running the list. Across several locations, that invisible list compounds into missed revenue every October through December.

Relay's AI employee cross-references the EHR due-date and lapsed-patient list against the schedule, identifies patients with expiring benefits who have not booked, and drafts personalized outreach through your existing comms tool. A staff member confirms the list before it sends. No-show slots are queued to a waitlist for one-click fill. Outreach never sends autonomously; your team finalizes.

Benefit-year-end creates a use-it-or-lose-it demand spike for calendar-year vision plans every October through December that under-staffed groups under-capture every year. No-shows on premium slots are disproportionately costly, and unused benefit-year capacity simply walks out the door.

The benefit-year recall list gets worked before the window closes, not after it resets.

Incomplete intake pushes eligibility to check-in, slows chair turnover, and surfaces dual-track errors as denials on the back end instead of catching them at the front desk.

Relay's AI employee sends a pre-visit intake sequence, collects member IDs for both vision and medical coverage, runs a background eligibility check on both tracks, surfaces any mismatch before the appointment, and reminds when the packet is incomplete 48 hours out. A staff member finalizes the record in the practice management system before the patient arrives. It works on top of RevolutionEHR, Eyefinity, and your existing intake tool, so the data lands once.

Optometry intake is two intakes. Staff need the vision-plan member ID, which is often different from the medical card, the benefit year, and whether the patient carries both vision and medical coverage (diabetes, dry eye, glaucoma). Pediatric optometry adds parental consent, school vision-screening referrals, and Medicaid EPSDT documentation. At a multi-location group, the same patient often gets re-entered from scratch at a second site because the record does not follow them.

Dual-track intake errors get caught at the front end, before they become back-end denials.

The most common failure mode for active providers is a missed CAQH attestation, required every 120 days, which can flip a provider inactive and trigger mid-stream denials no one expected.

Relay's AI employee maintains a credentialing tracker across every OD and payer panel: license and DEA expirations, CAQH attestation cycles, and payer re-enrollment windows, with a renewal calendar that alerts the coordinator ahead of each deadline. It drafts renewal and new-hire application packets and tracks submission status. A staff member signs off; the AI does not submit on its own.

Each new location must credential its ODs with each payer individually, on two parallel tracks: medical payer enrollment (Medicare, Medicaid, commercial) and vision-plan credentialing (VSP, EyeMed, Davis Vision, Spectera). Credentialing cycles run long, and a new hire at a satellite cannot bill under their own NPI until it is complete. Decentralized credentialing across sites creates inconsistent processes and enrollment gaps that quietly hold revenue.

Credentialing lapses get flagged before they block payment, not after claims start rejecting.

RevolutionEHR is the dominant cloud-native EHR for independent and multi-location optometry groups, with the strongest adjacent-tool ecosystem in the category. Relay's AI employees work on top of it through its gated FHIR R4 access (SMART on FHIR, OAuth 2.0; third-party access requires partner approval), alongside the integrations the group already runs: ABB Verify for eligibility, RevClear and TriZetto for RCM, Ensora and APEX EDI as clearinghouse, Weave or Solutionreach for comms, and IntakeQ for intake. Relay reads the data RevolutionEHR holds and works the eligibility, auth, denial, recall, and AR-reconciliation queues. Relay does not replace RevolutionEHR; a staff member finalizes everything the AI prepares.

Eyefinity, a VSP company, has the largest installed base among established multi-doctor and small-group practices: OfficeMate plus ExamWRITER historically, now unified as Encompass. Relay's AI employees work on top of it through Eyefinity's gated FHIR access, alongside native VSP eligibility and authorization and TriZetto ERA processing. ExamWRITER's ONC certification will be withdrawn in May 2027 and 2026 is the last MIPS reporting year for ExamWRITER users, so groups are evaluating their downstream automation now. Relay layers on top; it is not a replacement.

Compulink Advantage is the enterprise-tier platform for 4-plus-location groups and DSOs, covering clinical, optical retail, vision and medical billing, and analytics. Relay's AI employees work on top of its published FHIR API and HL7 support to close the gaps operators report most: a ledger that struggles across multiple locations and ERA handling that requires more manual intervention than most groups expect. Relay reads claim, schedule, and auth data and surfaces the next action, and a staff member finalizes; Relay layers on top rather than replacing the platform.

MaximEyes (First Insight) serves independent and regional group practices; ModMed EMA serves ophthalmology practices and medical-model optometrists co-managing surgical patients. Both hold ONC certification and expose FHIR APIs through that certification. Relay's AI employees work on top of either through that FHIR surface to run the downstream eligibility, billing, and denial cycle. A staff member finalizes every action.

Nextech is used by medical-model and ophthalmology-adjacent optometry practices. Relay's AI employees work on top of Nextech's FHIR surface for the same downstream eligibility, billing, and denial cycle. A staff member finalizes every action.

Relay's AI employees also work alongside the point tools your group already runs: pVerify, Availity, and ABB Verify for eligibility; Waystar, TriZetto, Office Ally, and Ensora and APEX EDI for clearinghouse and RCM; Weave, Solutionreach, and NexHealth for patient comms; Phreesia, IntakeQ, and Lobbie for intake; Updox and Doximity DocFax for HIPAA-compliant fax; CAQH ProView, Medallion, and Modio Health for credentialing. Relay does not replace these tools; it orchestrates across them so the data and the next action land in one place.

They run both tracks simultaneously. The AI verifies vision-plan and medical coverage separately before the visit, checks that each claim's service type maps to the correct payer, and flags the 920xx-vs-992xx coding split before submission. A staff member reviews and releases. Relay works on top of the EHR you already run.

No. Relay is not an EHR. The AI employees work on top of the platform your group already runs, through its FHIR access, and a staff member finalizes every action. You keep your system of record.

Yes. CMS-0057-F decision-timeframe requirements take effect January 1, 2026, cutting the standard prior-auth decision window to 7 calendar days (the separate Prior Authorization API requirement follows on January 1, 2027). The AI employee monitors every open medical-eye auth against that window, alerts staff before the visit date, and drafts the submission packet for one-click review.

They triage denials by root cause across both payer tracks, draft the correction or appeal for common patterns, and surface denial trends by payer, code, and location so the worst-performing site stops being invisible. Your billing coordinator approves and submits.

The AI does the bulk of the work, watching queues, pulling data, drafting the action, and a staff member reviews and finalizes before anything is submitted or sent. The AI gates the queue; it does not act autonomously.

Yes. The AI runs a nightly sweep across both tracks for everyone scheduled in the next 48 to 72 hours, including Medicare Advantage vision carve-outs, and surfaces a worklist of exceptions for staff to resolve before the patient arrives.

Yes. The AI cross-references the EHR due-date and lapsed-patient list against the schedule, finds patients with expiring benefits who have not rebooked, and drafts outreach through your existing comms tool. A staff member confirms the list before it sends.

The intake AI employee handles pre-visit data capture, runs a live eligibility check against the patient's stated insurers, and offers to queue a booking, escalating clinical questions to a staff member. It works on top of your scheduling and intake tools and everything it produces is reviewed by staff before it reaches the patient.

Through Eyefinity's gated FHIR access, alongside native VSP eligibility and TriZetto ERA processing. With ExamWRITER's ONC certification ending in May 2027, groups are evaluating downstream automation now; Relay layers on top without a rip-and-replace.

No. Doctora and other scribes own the exam room. Relay handles everything downstream of the signed note: eligibility, prior auth, billing, denials, recall, and credentialing. The two are additive.

Yes. The AI maintains a credentialing tracker per OD and payer panel, including CAQH attestation cycles required every 120 days, license and DEA expirations, and re-enrollment windows, and alerts the coordinator before each deadline. Staff signs off on every submission.

Multi-location groups, roughly 2 to 20 sites, where the dual-payer admin load scales with patient volume but the back-office headcount does not. The leak grows every month you run more sites on the same team.

The first two to three weeks are discovery: team working sessions, mapping the operation, and building the first AI employee. Builds run 8 to 12 weeks. There is no rip-and-replace of the EHR; setup runs on top of the platforms you already have.

Yes. Relay operates under a Business Associate Agreement. The AI employees access only the data they need to run the workflows they are built for, through the FHIR interfaces your EHR already exposes. Ask on the intro call for the BAA and security documentation.

Headcount scales linearly with locations; the cross-location visibility gap does not close by adding staff. The AI employees watch every location's queues simultaneously and surface the action for a staff member to finalize. It is not a replacement for your team; it is the orchestration layer that keeps the team from drowning in volume.