Skip to content
be-human.ai
← Back to the journal

AI by industry

Superintelligence for Ophthalmology Practices: Surgery Scheduling Admin

How an ophthalmology practice can hand surgery scheduling, pre-op checklists and insurance follow-up to AI agents while surgeons keep every clinical decision.

Be Human typographic cover with the title: Superintelligence for Ophthalmology Practices: Surgery Scheduling Admin
Conceptual illustration created with AI · Be Human

Superintelligence for an ophthalmology practice means AI agents that take over the administrative work around eye surgery, including the surgical calendar, pre-operative checklists and insurance follow-up, while each agent works from a written memory, a short list of tools and clear approval rules. The surgeon keeps every clinical decision, from candidacy and procedure choice to how a finding changes timing and what a patient is told about risk. The practical question for an operations leader is not whether an agent can write a reminder, but which steps it may complete alone, which it may only prepare, and which must always wait for a named person.

Key takeaways

  • Hand the surgical calendar, pre-op checklist delivery and insurance status tracking to agents first, because these tasks are repetitive and easy to verify.
  • Keep candidacy, diagnosis, timing changes, consent content and any answer about symptoms or risk with the surgeon and clinical staff.
  • Give each agent a written memory of approved templates and rules, a short tool list and an explicit approval boundary before it touches a patient or payer.
  • Measure the agent with countable outcomes, such as cases that reached surgery day with a complete checklist, reviewed on a fixed schedule.
  • Start with one role and one workflow, and expand only after the first one has run under review and its mistakes have been fixed in the written rules.

Where administrative work ends and clinical judgment begins

Eye surgery involves more people than most patients see. The surgeon makes the clinical call, but a surgical coordinator books the case, a front desk scheduler holds the room, pre-testing staff collect results, a biller follows the authorization, and sometimes an outside surgery center or hospital controls the operating room schedule. Most delays happen in the handoffs between these people, and most handoff work is information moving from one record to another, a deadline being remembered, or a patient being asked for a document. That is the work an agent can carry. The line between administrative and clinical is easiest to hold when a decision guide is written down and signed by the surgeon before any agent goes live.

  • Agent may complete alone: confirmation messages using approved wording, document requests, reminders for deadlines the surgeon has already set, and status updates copied from the chart.
  • Agent may prepare, staff must approve: proposed calendar holds, rescheduling options inside open block time, and insurance packets assembled from charted data.
  • Surgeon only: candidacy, timing changes after a pre-op finding, interpretation of test results, any change to consent content, and any answer to a patient describing symptoms or risks.

How do you build a surgery calendar agent without double bookings?

Surgical calendars carry rules that do not show in the booking grid: each surgeon's block days, case lengths by procedure, room and equipment needs, anesthesia requirements, and which cases must follow a completed pre-op visit. An agent can apply these rules only if they are written down. Start by turning the habits of the surgeon and coordinator into a short scheduling document, then give the agent read access to the calendar and write access only to a queue of proposed holds.

1. Write the scheduling rules with the surgeon and coordinator: block owners, case lengths, minimum lead times, and which procedures need a completed pre-op visit first. 2. Give the agent read access to the calendar and permission to create proposed holds, not confirmed bookings. 3. Require the agent to list every conflict it finds, and forbid it from resolving a conflict by moving another patient. 4. Require a named staff member to confirm each hold before it becomes a booking and before the patient receives any confirmation. 5. Review the proposal log every week, and change the written rules when a proposal was wrong instead of adjusting the agent's instructions informally.

Illustrative scenario: A practice with two surgeons and one shared outpatient room receives a cancellation late on a weekday afternoon. The agent finds three waiting-list patients whose procedures fit the freed time, checks which of them have a completed pre-op visit, and posts a ranked proposal to the coordinator. The coordinator calls the first patient and confirms. Only then does the agent update the calendar and send the approved confirmation, and the chart stays unchanged until a person signs off.

What belongs on a pre-op checklist, and who reads the answers?

A pre-operative checklist is long, and patients skip items they do not understand. An agent can send the checklist in the patient's preferred language, track which items came back, and send one reminder per open item on a schedule the practice approved. It should not interpret the answers. When a patient reports a new symptom, names a medication the surgeon has flagged, or asks whether a sensation is normal, the agent stops, tells the patient that a member of the clinical team will reply, and creates a same-day task with the message attached.

  • Procedure, eye and date confirmed against the surgeon's order
  • Pre-op visit and required tests completed and marked as reviewed by the surgeon
  • Medication list updated, with any flagged medication routed to the surgeon
  • Fasting and arrival instructions sent in the patient's preferred language
  • Escort and transport confirmed for the day of surgery
  • Consent forms received and signed before the day of surgery
  • Insurance authorization or benefit check recorded in the chart
  • Any question about symptoms or risks routed to clinical staff the same day

How should insurance follow-up work without guessing at coverage?

Insurance follow-up for eye surgery covers benefit checks, prior authorization requests, claim status and denial letters. An agent is useful for the repetitive parts: gathering the chart notes and test results a payer asks for, filling the packet from data the clinical team has already recorded, and tracking each request against its deadline. The agent should not choose billing codes, argue medical necessity, or tell a patient that coverage is certain. Denials go to billing staff with the payer's reason text attached, and appeals stay with people who know each payer. Teams that have worked through prior authorization at imaging centers will recognize the same division of work. Denials that do not resolve with a simple resubmission follow the pattern in medical billing denial follow-up.

Permissions, patient data and prompt injection

Patient messages are untrusted text. A message can contain instructions, written by mistake or on purpose, and an agent that treats that text as a command can be steered into the wrong action. The OWASP guidance on prompt injection explains why instructions and data must be kept apart. In practice, give each agent only the fields and actions its role requires, use scoped access to the calendar and messaging systems, and log every action with the record it read and the rule it applied. Keep patient health information inside systems the practice already governs. The NIST AI Risk Management Framework gives a useful vocabulary for listing these risks before launch, and privacy obligations depend on where your patients live, so review them with your own advisors.

How Be Human's managed approach handles this

Be Human designs and operates this kind of setup as a managed service. The first engagement defines one useful role, such as a surgery scheduling coordinator agent, and writes down the memory it needs, including approved templates and scheduling rules, the tools it may call, the approval boundaries that match the surgeon's decision guide, and one observable success criterion. Ongoing operations can then include monitoring of the agent's actions, recurring evaluations against past cases, maintenance when the calendar or messaging integration changes, periodic permission review, incident response, and controlled improvements. The work is remote and available in English and Portuguese.

How do you know the agent is working?

Success criteria should be countable and tied to the surgeon's rules. Useful examples include the share of booked cases that reached surgery day with a complete checklist, the number of clinical questions escalated on the same day, the number of proposed holds that staff rejected and the reasons, and how many insurance requests left the practice before their deadline. Log each action so a reviewer can reconstruct what the agent read and why it acted, which is the same discipline the OpenTelemetry observability concepts describe for software systems in general. Run a fixed set of past scheduling and checklist cases through the agent at regular intervals, so that changes to the model, the tools or the rules show up before any patient is affected.

If you want to find the first surgery admin task worth handing to an agent, take the free Be Human AI diagnostic.

Frequently asked questions

Can an AI agent book eye surgery on its own? No, not in the sense of confirming a surgical date without a person. The agent can read the calendar, propose holds and prepare the booking, but a named staff member confirms each one. That keeps the surgeon's block rules and room constraints in human hands.

Who decides what a patient receives before surgery? The surgeon and practice decide the content. They approve the checklist, the fasting instructions and the message templates in writing, and the agent sends only those approved texts. Anything outside a template goes to clinical staff.

What should happen when a patient asks a clinical question through the agent? The agent should not answer it. It tells the patient that a member of the clinical team will reply, records the question, and creates a same-day task with the patient's message attached, so the reply comes from a person who can see the chart.

Does this replace our billing or surgical coordinators? Usually not. The agent takes over repetitive steps such as assembling packets, tracking deadlines and sending status requests, so coordinators and billing staff can spend more of their time on exceptions, appeals and patient calls. Staffing decisions stay with the practice.

Your next step

See where an AI agent fits in your company first.

Take the free Be Human AI diagnostic. A few questions about your work show which task to hand to an agent first, what it needs to know and where a person approves.

Be Human editorial guidance. Examples are illustrative unless explicitly attributed. Sources provide background; implementation choices depend on your workflow.

Less busywork. More room for you.

Bring us the work you wish you had time for.

Book a call to try a personalized AI agent free for 7 days ↗Or take the free AI diagnostic →