The consult bills. The physician does it.
The visit where the decision to operate is made is billable outside the global. Nothing the agent does before it is a service Medicare pays for.
Almost everything the care line does sits inside cataract surgery's 90-day global period, which Medicare already pays for, and no CPT code pays for an AI's time. The money is the surgery itself: the case that cancels on the day, the second eye that never gets booked, and the after-hours nurse time the question line absorbs. Billing codes exist after day 90, for a separate chronic condition and on human staff time. B+L bills none of it, and should not want to.
The visit where the decision to operate is made is billable outside the global. Nothing the agent does before it is a service Medicare pays for.
No code pays for an AI call. A nurse's prep call is unbilled overhead today, which makes it a cost the agent can take on, not revenue.
Post-op visits, calls and complications that don't need the OR are bundled (about 20% of the fee is the post-op share). Remote monitoring by the practitioner who holds the global can't be billed for surgery-related care, and virtual check-ins must be done by the clinician personally. Co-management (-54 / -55) splits the same bundled fee; it adds nothing.
Only for a separate chronic condition (glaucoma likely; dry eye unlikely to meet PCM's risk bar), with consent, a care plan and human clinical-staff minutes. RTM needs an FDA-defined device, and claiming one would undercut the agent's non-device status.
The booking and before-consult calls keep the first visit on the calendar. Not measured yet in the rig.
About 4% of ophthalmic ASC cases (2011 survey), about 40% of them facility-initiated for medical or prep reasons. Surgery week targets the prep failures.
≈38% of 2025 US cases include a premium component. The agent gathers and never recommends; B+L making IOLs is why that line is absolute.
RN wage cost of one 10-minute call (BLS median, before benefits). The question line answers from protocol and triages the rest.
≈2.8% of patients book one in month one. Caught earlier or answered by phone; the rig doesn't measure it yet.
Another full case. Framed as time-to-second-eye for patients already listed, never as a sales call.
Every code someone will suggest, with the reason it does or doesn't fit. National 2026 Medicare rates are from a third-party lookup of the fee schedule [5]; check them against CMS's PFS lookup before they go in a deck.
| Code | What it pays for | Fits? | 2026 national | Why, in one line |
|---|---|---|---|---|
| 66984 | Cataract surgery, 90-day global | the case | $1,255.73 ASC + $462.94 surgeon | Pre-op visits after the decision, all post-op follow-up and non-OR complications are inside it [1][3][4]. |
| 99424 – 99427 | Principal Care Management | conditional | $87.51 / $61.46 / $67.80 / $54.11 | One high-risk condition (glaucoma possibly; dry eye unlikely), 30 human minutes a month, consent and a care plan [12]. |
| 99490 / 99439 | Chronic Care Management | conditional | $66.13 / $50.44 | Two or more chronic conditions, 20 minutes of clinical staff time, consent, a comprehensive care plan [12]. |
| 98975, 98979 – 98981 | Remote Therapeutic Monitoring: setup, management | conditional | $21.71 · $26.39 / $54.11 / $41.42 | Needs an FDA-defined device; adherence data qualifies; never for surgery-related care inside the global [6][7][9]. No eye supply code exists. |
| 99453 – 99458 | Remote Physiologic Monitoring | no | $21.71 – $52.11 | Needs physiologic device data; answers to questions aren't physiologic [9]. |
| 98016 / G2252 | Virtual check-in | no | $17.37 / $28.39 | Patient-initiated and performed by the clinician personally; 98016 replaced G2012 in 2025 [13]. |
| 99421 – 99423 | Online digital E/M | no | $15.70 / $30.73 / — | Counts the clinician's own time, not an agent's. |
| 99211 | Minimal office visit | no | $24.38 | Face-to-face clinical staff; bundled inside the global [1]. |
| MIPS IA | Improvement Activities credit | indirect | no dollars | 2026 adds an AI patient-safety activity; whether after-hours access counts is unverified [34]. |
On AI time. CPT defines clinical staff as a person working under supervision, so agent minutes can't count toward any time-based code [15]. The AMA's AI taxonomy (Appendix S, revised June 2026) classifies AI output, not billable time [14], and no Category I or III code pays for AI patient communication. The CY2027 proposed rule asks for comment on "technology-enabled care management" codes and would limit RPM/RTM to practice-employed staff [10][11]. Watch that rule; don't build a plan on it.
The business case is the ≈$1,719 case kept on the schedule, the second eye booked, and nurse hours returned. Say it that way, and nobody in the room has to defend a code that doesn't exist.
Day-of-surgery cancellation rate, time-to-second-eye, and after-hours pages avoided, reported only alongside triage sensitivity and red-to-person minutes. Each needs a baseline from the first practices; none is measured yet.
Kickback, Open Payments and beneficiary-inducement questions come before any chronic-care billing pilot. A jury found an IOL distributor paid kickbacks to cataract surgeons, settled in 2024 [21]; that is the precedent in the room.
Answered where the public record answers it; marked open where it needs a number from a practice or a call from counsel.
The practice: it's the enrolled billing provider and the surgical case is its revenue. B+L captures none of it directly. That's by design; the reason is under "What has to be signed off".
No. B+L isn't an enrolled Part B practitioner, and the CY2027 proposal would require practice-employed staff even for remote monitoring [10]. B+L's return is strategic: practice loyalty, data on the journey, and the regulated product on the build-out.
Case volume × cancellation rate × the share of cancellations that are prep failures, plus second-eye timing and after-hours nurse hours. Illustration only: 1,000 cases, a 4% cancellation rate and a quarter of those saved is ten cases, ≈$17,000 of Medicare revenue kept, before second eyes and staff time. The first pilots have to produce the baseline.
After day 90, for glaucoma or another separate chronic condition, on human nurse minutes: PCM at $67.80 a month for 100 patients is ≈$81,000 gross a year, before coinsurance, and it needs 30 nurse minutes per patient per month. The agent can prepare and route that work; it can't be the minutes.
The Anti-Kickback structure (fair-market-value fees or no billable output), whether the platform is an Open Payments transfer of value, beneficiary-inducement limits ($15 an item, $75 a year [20]), TCPA consent for AI voice [23], state AI-disclosure laws such as California AB 3030 [24]. An OIG advisory opinion is an option: one in March 2026 favoured a brand-agnostic ophthalmic software arrangement the practice paid for [17].
It can. Medication reminders and symptom questionnaires sit under FDA enforcement discretion today [25][26], and RTM billing requires an FDA-defined device. Billing RTM would argue the agent is a device, against the positioning the regulated track depends on. Needs regulatory counsel before anyone says "RTM".
Not inside the global: that's post-op care, already paid for in the surgery fee [1]. The value is the nurse's hour spent on the calls that need her instead of the ones that don't.
Patients may be billed the premium portion under CMS rulings, and ≈38% of 2025 US cases had a premium component [29][30]. The agent never recommends or ranks a lens; a manufacturer-funded agent that did would be the fastest way to lose the program. Upside, if any, comes from a better-prepared consult, not a nudge.
Not from a US code: its December 2025 US launch at Dell Medical School names no CPT or reimbursement model [33]. Nobody has a code. The advantage goes to whoever proves throughput and cost first.
The CY2027 proposed rule asked for comment on technology-enabled care management; comments closed 14 September 2026 [10]. If a code arrives, the same kickback question applies to it. Track the final rule (expected November) and don't count on it.