How an open slot becomes a booked appointment
An athenaOne schedule change starts one operating loop: classify the opening, select the permitted patient cohort, commit the schedule transaction, and log the outcome. Pretty Good AI reads the current athenaOne state again before booking, so an offer cannot rely on a schedule snapshot that has already changed.
-
Detect the trigger. A slot opens, a patient cancels without rebooking, an earlier appointment becomes available, or a provider blocks scheduled time.
-
Read the live schedule. The agent retrieves slot availability, appointment-type eligibility, provider templates, departments, existing appointments, and the athenaOne waitlist.
-
Build the candidate pool. Patients are ordered under the practice's waitlist, acceleration, notice-period, and contact rules.
-
Apply the eligibility gate. Appointment duration, payer enrollment, and authorization status must pass before the patient receives an offer.
-
Contact eligible patients. Offers go out by voice or text. Unanswered calls follow the configured retry sequence while the agent continues working the eligible list.
-
Commit the accepted offer. The agent confirms that the slot remains open, books or moves the appointment in athenaOne during the interaction, and stops outstanding outreach for that opening.
-
Route exceptions. Any failed eligibility test or unresolved scheduling condition goes to a scheduler with the specific reason attached.
| Pretty Good AI runs these workflows as connected branches of one capacity-recovery sequence. The broader workflow inventory is available in the athenaOne workflow catalog. | |||
| Workflow branch | Trigger | Patient cohort | athenaOne result |
|---|---|---|---|
| Waitlist backfill | A matching slot opens. | Eligible patients on the athenaOne waitlist, worked in the practice's approved order. | The accepting patient is booked into the open slot; unanswered patients follow the retry rule. |
| Cancellation recovery | A patient cancels without choosing another time. | The cancelling patient plus eligible waitlist patients for the released slot. | The original patient receives matching alternatives while the freed slot enters waitlist backfill in parallel. |
| Appointment acceleration | An earlier opening can accommodate a patient already booked later. | Booked patients who meet provider, visit-type, notice, acceleration, payer, and authorization rules. | The patient moves forward and the later appointment becomes available for another patient. |
| Auto-rescheduling | A provider blocks days, changes availability, or leaves the practice. | Patients whose appointments are affected by the schedule change. | The agent rebooks by call or text and retains the campaign context when a patient calls back. |
athenaOne data used for each decision
Pretty Good AI has 820+ athenaOne APIs in production. Capacity recovery specifically uses live scheduling objects rather than a copied calendar or middleware queue. Examples include GET /v1/{practiceid}/appointments/open, GET /v1/{practiceid}/appointmenttypes, and GET /v1/{practiceid}/appointments/waitlist.
| Input | How the workflow uses it |
|---|---|
| Live slot availability | Confirms that an opening still exists before outreach and immediately before booking. |
| Appointment types and durations | Limits offers to visits that fit the slot's configured type and length. |
| Provider templates | Preserves provider-specific restrictions, session structure, and visit placement rules. |
| Departments | Keeps the proposed visit at an eligible location and under the correct departmental configuration. |
| athenaOne waitlist | Supplies the active patient pool and the ordering rules used for waitlist outreach. |
| Existing appointments | Identifies patients who can move forward and prevents duplicate or conflicting bookings. |
| Payer and authorization state | Prevents an administratively ineligible patient from receiving an opening that staff would later need to unwind. |
Accepted offers become schedule transactions inside athenaOne. Staff see the resulting appointment in the same schedule they already operate, with no separate booking queue to reconcile. athenaOne exposes appointment records and scheduling actions through its appointment APIs; the distinction between direct writeback and middleware handoff is covered separately in Native EHR Writeback vs Middleware in Healthcare Voice AI.
The eligibility gate before an offer goes out
Every candidate must pass three administrative tests before Pretty Good AI offers the slot. A failure creates a scheduler handoff with the failed condition attached; the agent does not make the booking and leave staff to reverse it later.
| Eligibility test | Pass condition | Failure handling |
|---|---|---|
| Appointment type and duration | The patient's required appointment type is permitted in the opening, and its configured duration fits the provider template. | The patient receives no offer. The scheduler receives the appointment type, required duration, offered slot, and mismatch reason. |
| Payer enrollment at the location | The proposed provider can see the patient's payer at the department or location attached to the opening. | The patient receives no offer. The handoff identifies the payer, provider, and location combination that failed. |
| Authorization constraints | No authorization is required, or the authorization has returned and remains valid for the proposed service date. | A visit with a pending authorization can be pushed later but cannot be pulled earlier. The scheduler receives the current appointment, proposed date, and authorization state. |
Authorization directionality matters most in appointment acceleration. Moving a pending case later preserves time for the authorization to return; moving it earlier can place the visit before approval. Pretty Good AI applies that constraint before patient contact, consistent with its authorization-window scheduling analysis.
Offers, declines, retries, and stale-list protection
Pretty Good AI uses voice as well as text because the available patient pool is wider than the group that reads and responds to SMS. The agent completes the booking during the accepting call or text thread; it does not ask the patient to call the front desk before the opening disappears.
| Patient or schedule event | System action |
|---|---|
| Patient accepts | Recheck the live slot, book or move the appointment in athenaOne, confirm the result, and stop competing offers. |
| Patient declines | Record the decline for this opening and continue to the next eligible candidate. Future waitlist eligibility follows the practice's configured rule. |
| Patient does not answer | Apply the configured retry cadence and continue working the ordered candidate list instead of holding the opening indefinitely. |
| Patient calls back | Recognize the active waitlist or rescheduling context, recheck availability, and resume the transaction without making the patient repeat the reason for the call. |
| Patient is already booked | Skip the stale waitlist entry and record the reason. No new offer is sent. |
| Slot has been taken or closed | Stop outreach for that opening. Patients with active scheduling needs remain available for later matching slots under practice rules. |
| Eligibility changes before acceptance | Cancel the proposed transaction and route the case to a scheduler with the failed gate identified. |
Illustrative scenario: recovering a slot with 51 hours' notice
A patient cancels a Thursday 1:00 p.m. follow-up appointment on Tuesday at 10:00 a.m. without rebooking. The 51-hour opening qualifies for short-notice recovery, and cancellation recovery starts for the original patient while waitlist backfill starts for the released slot.
-
The athenaOne waitlist contains 18 active entries associated with the department.
-
Seven entries fail the appointment-type or duration test, three fail the payer and provider-location test, and two have pending authorizations that prevent an earlier move.
-
Six patients remain eligible. The first patient does not answer, the second declines, and the third accepts 12 minutes after the opening is detected.
-
Pretty Good AI rechecks the Thursday slot, books the third patient in athenaOne, confirms the appointment, and stops outreach to the remaining candidates.
-
The original patient accepts a later matching appointment through the cancellation-recovery branch. Both schedule changes are visible in athenaOne without staff re-entry.
The Thursday opening counts as refilled at booking, but it becomes one net recovered visit only after the replacement appointment is completed. A later cancellation, no-show, or denied visit remains attached to the fill so booked activity is not mistaken for recovered capacity or revenue.
Measuring capacity and revenue actually recovered
The measurement unit is the slot that opened, not the message, call, or appointment record produced during outreach. Pretty Good AI separates initial fills from visits that later cancel, no-show, or produce a denial.
| Metric | Calculation | What it establishes |
|---|---|---|
| Slot utilization | Completed visits divided by published bookable provider slots for the measurement period. | Whether capacity that the practice made available was ultimately used. |
| Short-notice refill rate | Slots opened within 72 hours and refilled divided by all slots that opened within 72 hours. | Whether the workflow recovered openings that staff had limited time to fill. |
| Refilled visits later lost | Replacement appointments that later cancel or no-show divided by all refilled appointments. | Whether the initial booking survived through the scheduled visit. |
| Net recovered visits | Completed replacement visits, excluding later cancellations, no-shows, and visits removed from the recoverable set under the practice's reporting rules. | The number of patient visits attributable to recovered openings. |
| Denied refilled visits | Completed replacement visits that later produce a denial, tracked separately by payer and denial reason. | Whether a filled chair also produced a viable revenue-cycle outcome. |
| Recovered revenue | Net payments from completed recovered visits. Allowed amounts can serve as a provisional value until payment posts, provided the same basis is used throughout the report. | The financial value that survived scheduling, attendance, claim processing, and payment. |
Messages sent, patients contacted, offers accepted, and appointments booked remain useful operating metrics, but none is recovered revenue. The 72-hour short-notice denominator and slot-utilization method are also defined in Pretty Good AI's capacity-recovery measurement framework.
Operating responsibilities
| Pretty Good AI handles | The practice provides and approves |
|---|---|
| Reading live athenaOne schedule, appointment-type, department, provider, waitlist, and appointment data. | Provider and appointment-type matching rules, including duration and template exceptions. |
| Applying the eligibility gate before patient outreach. | Payer enrollment rules by provider and location. |
| Running voice and text outreach, retries, callback recognition, booking, and writeback. | Notice periods, acceleration limits, authorization windows, and outreach priorities. |
| Stopping stale or competing offers after a schedule change. | The scheduler queues and owners who receive exceptions. |
| Attaching the failed condition when a candidate requires staff review. | Validation of representative appointment scenarios before launch. |
Frequently asked questions
Can Pretty Good AI fill a same-day cancellation without sending the patient back to the front desk?
Yes. Pretty Good AI reads the live opening, applies the practice's appointment, payer, and authorization rules, contacts eligible patients, and books the accepting patient directly in athenaOne. The agent performs a final availability check before committing the appointment, so the patient does not receive instructions to call the office and compete for the slot.
What happens if two patients try to accept the same opening?
The first successful athenaOne booking closes the opening. Pretty Good AI rechecks slot availability before each commitment and stops outstanding outreach after the schedule transaction succeeds. A later response is evaluated against the current schedule, so the agent can offer another eligible time or retain the patient's waitlist status instead of double-booking the original slot.
Can a patient with a pending authorization be moved into an earlier appointment?
No. Pretty Good AI does not accelerate a visit whose required authorization has not returned. The visit can be moved later when that follows the practice's rule, but an earlier offer fails the authorization gate and routes to a scheduler with the current date, proposed date, and authorization state attached.
How does auto-rescheduling handle a patient who calls back?
Pretty Good AI recognizes the patient and the active provider-block or rescheduling workflow when the return call arrives. The agent checks the current schedule, resumes with eligible replacement times, and writes the accepted appointment into athenaOne. The patient does not have to reconstruct the cancellation notice for a front-office employee.
Does Pretty Good AI have experience handling more than 100,000 patient calls a month?
Yes. Commonwealth Pain & Spine reports that Pretty Good AI handles more than 100,000 patient calls per month across 35 locations and 130 providers, with one in eight bookings occurring after hours. These are customer-reported operating figures that establish high-volume voice and scheduling use; they are not a published waitlist-backfill recovery percentage. Pretty Good AI customer deployments.