An AI appointment setter for real estate should be evaluated on what its “booked” label actually establishes. Before buying, ask it to distinguish an agent consultation from a property showing, a calendar record from a participant response, and a showing request from permission to visit. Require evidence for each claim before treating an appointment as ready.
This is a documentation-based buyer's acceptance rubric, not a product ranking or a test of any account. Callion publishes it and has a commercial interest in lead-response software. No Callion calendar or showing integration is asserted here.
A calendar record is only one part of the answer
Google Calendar documents attendee response states separately, including not responded, tentative, declined, and accepted. Google Calendar Events reference
Use that distinction to ask a more precise demo question: “What evidence supports the response label, and which system or person wrote it?” Google's response field is writable, so do not treat an accepted value alone as independent proof that a person clicked an invitation. Ask the provider to explain provenance and show a controlled example rather than merely displaying a green status. Google Calendar response field
For a consultation, specify the intended agent, participant, purpose and agreed time. For a showing, ask a separate question about access. A buyer agreeing to meet is not the same operational decision as a property being available for that visit.
Showing permission needs its own evidence
ShowingTime distinguishes Go & Show listings from Appointment Required listings; for the latter, confirmation must come from the seller, tenant, and/or listing agent before the showing may occur. Home by ShowingTime's notification guidance
That is a description of ShowingTime's documented workflow, not a universal legal rule or proof that a proposed AI product connects to it. In your evaluation, ask which showing process applies and who can verify its result. A provider that creates an agent calendar event should not receive credit for confirming property access unless it demonstrates that separate step.
Do not require a fresh seller response for every listing indiscriminately: the documented Go & Show distinction is precisely why a single “approval required” assumption is inadequate. Ask the responsible agent to establish the applicable process.
A buyer's appointment-evidence matrix
Use this proposed matrix to agree on acceptance criteria before a demo. These are requirements to investigate, not features attributed to any vendor.
| State to evaluate | Evidence to request | What the assistant may reasonably report in the test |
|---|---|---|
| Consultation requested | Participant's stated purpose and proposed time | “You requested a consultation.” |
| Calendar record created | The intended event and participants, with its actual status | “A calendar event has been created.” |
| Participant response recorded | Response value plus how it was obtained | A precise description of the recorded response, without inventing a human action |
| Showing requested | Property-specific request and the responsible agent | “The showing request is pending.” |
| Showing access confirmed | Applicable showing process and confirmation or authorization evidence checked by the responsible agent | A property-specific confirmation only after that evidence is established |
| Meeting held | A defined attendance record or agent confirmation | “Held,” rather than merely “booked” |
An absent record should produce “not established,” not an inferred pass. Have the vendor demonstrate a missing-evidence case and identify who receives the exception. If a provider only supports consultations, that can still meet your needs; evaluate it against that narrower purpose rather than crediting it with showing coordination.
Test whether changed details reach the participant
Calendly distinguishes calendar invitations from email confirmations; its documentation says the latter gives invitees a static event file rather than automatic updates to changed details. Calendly scheduling notifications
Treat this as a reason to test the chosen notification path, not a claim that Calendly never sends other change-related messages. Ask for a demonstration using an inbox and calendar your team controls: establish the initial details, change one detail through the agreed process, and inspect what the participant actually sees. Record the exact notification type and result. Do not assume the host's updated screen establishes the recipient's updated information.
A synthetic demo that exposes an ambiguous booking
Suppose a buyer and the intended agent have both agreed to the same Thursday consultation date and time, the calendar contains an event, and a recorded response says accepted. The same buyer also asked to see a property handled through an Appointment Required process, but the responsible agent has not established showing confirmation.
In this invented test, the consultation and the property request have different states. A suitable message would be: “Your consultation is scheduled. The property visit is still awaiting confirmation; the agent will update you.” “Your tour is confirmed” would overstate the evidence.
Next, have the responsible agent supply a simulated showing confirmation and ask the vendor to show exactly which record and participant message change. Finally, change the meeting location in the controlled test and inspect the recipient's resulting information. Save expected and observed outcomes separately. This is a proposed evaluation, not a report of tests performed by Callion.
Define success before comparing prices
Ask each shortlisted provider to state which outcome it counts: consultation calendar records, participant responses, authorized showings, or meetings actually held. Agree on how cancellations, duplicate records and unresolved requests will be treated in your own evaluation. Do not compare unlike counts as though they measured the same work.
This rubric does not establish legal permission to contact a lead, property-access rights, plan availability or any vendor's conversion performance. Account-specific setup and the brokerage's showing process still need verification.
For the implementation details of calendars and exception handling, see the lead-response automation guide. For the broader division of AI and human work, explore the AI ISA library. The decision here is narrower: can the provider show what each appointment label means, and stop short of claiming more than its evidence supports?