Built for GoHighLevel Voice AI

Your AI said it booked. Did it actually hit the calendar?

ActuallyBooked classifies each returned booking action and compares joinable actions with current calendar records—without taking control of your account.

Read-only Runs on your machine No recordings or transcripts in the report
Illustrative dataRead-only

Booking Integrity Report

4 bookings to checkLast 30 days · generated locally
98%matched
210Matched
2No match
2Ambiguous
0Needs data
StatusBookingRecordedReason
No match8f2a91c4Jul 12 · 14:08No calendar record in the window
Ambiguous31bda70eJul 09 · 18:41One record lines up with two actions

Hashed identifiers and exact times stay on the operator’s machine.

One narrow question, answered with evidence.

AIBooking actionWhat the agent recorded
31Calendar recordWhat exists right now

Check the evidence before you trust the outcome.

No dashboard migration, no account handoff, and no multi-week setup.

01

Run the free preflight

Use one Voice AI read permission to count booking actions with enough linking data. It reads no calendar data and does not persist GHL response data.

02

Reconcile the last 30 days

After payment, run the full audit locally with three view-only permissions. It compares each action with current calendar records.

03

Review what needs attention

Keep the detailed report local. Share only the summary counts, then review uncertain cases with us by email or a short walkthrough.

A mismatch is a lead to investigate—not a verdict.

ActuallyBooked compares recorded APPOINTMENT_BOOKING actions with the calendar records visible now. A missing match may mean a booking was never created, or that it was later canceled or moved.

It does not determine root cause, verify the caller’s requested time, or diagnose no-shows.

MatchedA matching calendar record exists.
!
No matching recordWorth checking case by case.
?
AmbiguousThe evidence does not support one clear match.
Needs more dataNot enough linking information to judge.

Your token never comes to us.

The checker runs on your computer. Your read-only token is sent over HTTPS only to GHL’s API, then you delete it when the audit is done.

Download the checksum
01

View-only permissionsThe audit cannot create, edit, cancel, or delete appointments.

02

No telemetry or uploadNo founder server receives your token or report.

03

Raw fields stay in memoryNames, phone numbers, recordings, and transcripts are not written to the report.

04

Detailed evidence stays localThe HTML contains pseudonymized operational data, so share summary counts only.

Start free. Pay only after the preflight confirms enough joinable actions to audit.

Booking Preflight

$0one GHL location

Counts structured booking actions from the last 30 days. No calendar access.

  • One Voice AI view permission
  • Runs locally
  • Source included
  • Requires Node.js 22+ · about 3 minutes
Download free preflight
First 3 pilot customers

30-Day Booking Audit

$99one-time · per location

Full local reconciliation plus a review of findings with the person who built the checker.

  • Three view-only permissions
  • Detailed local HTML report
  • Email or short walkthrough review
  • Refund if the audit cannot technically complete on the documented supported setup
Request the $99 audit

Pilot excludes healthcare, dental, financial, and legal accounts while the privacy and compliance program is still being built.

Built by an Imperial Computing graduate with company operating experience.

Your first audit is not routed into a generic support queue. The person building the system reviews the evidence and learns from every labeled finding.

Ask the founder

Before you run anything.

Does “no match” prove the AI failed?

No. It means no current calendar record matched the booking action under the audit rules. The appointment may have been canceled or moved later, so each finding needs human review.

Does the audit verify the date and time the caller requested?

No. It verifies whether the recorded booking action lines up with a calendar record created around that action. Requested-time correctness is a separate problem.

Can ActuallyBooked change anything in GHL?

No. The token uses view-only permissions. The customer-side checker contains no create, update, delete, cancel, or reschedule operation.

What does the 100-action preflight threshold mean?

It is the current pilot threshold for having enough structured actions to make a broad audit worthwhile—not a claim about the expected mismatch rate. If you already saw a booking that did not land, you may still qualify below 100.

Don’t ask whether your AI sounded convincing. Check whether the calendar agrees.