public-safe proof page
The studio assistant that turns live client SMS into safe operational decisions. This page uses fake, redacted examples only: no real clients, private transcripts, phone numbers, or screenshots.
moat snapshot
Inbound
[REDACTED_CLIENT] asks for a full evening class.
State
Class is full, waitlist open, active pack found.
Guardrail
Offer waitlist. Do not consume credit yet.
Reply
Confirm waitlist and explain the next step.
Booking requests, package questions, waitlist asks, and late-cancel language resolve into clear intents.
Capacity, credits, upcoming bookings, and policy windows are checked before Becky drafts a reply.
Charged changes, unclear requests, and package purchases stay behind confirmation or owner review.
live SMS sandbox
This sandbox uses fixed fake data and a local rules engine. It shows the moat: Becky reads intent, checks studio state, applies guardrails, and writes an owner-readable trail.
Sandbox run
Fill a late opening
Hi, can I get into the 6:00 PM reformer class tonight?
received
Class lookup: 6:00 PM is full, waitlist has one open slot, package balance is active.
logged
I can add [REDACTED_CLIENT] to the 6:00 PM waitlist. If a mat opens, I will confirm before booking the credit.
drafted
Yes please.
received
You are on the waitlist for 6:00 PM. I will text you if a spot opens and hold the booking for confirmation.
sent
Intent
highClient wants a new booking, not a schedule question.
Eligibility
highClient has an active pack, so Becky can offer the waitlist path.
Guardrail
highFull class means waitlist first; no credit is consumed without confirmation.
[REDACTED_CLIENT] asked for a full class, accepted waitlist, and Becky logged the hold without changing payment state.
Next action: Watch for cancellation, then ask for confirmation before booking.
Type a redacted example, then run the same deterministic classifier used by the sandbox.
redacted decision-log reel
The full reel can be produced from these beats. The current public page ships the script, visual sequence, and redacted evidence trail without exposing private transcripts.
10 seconds
Visual
Split view of a class schedule, unread SMS bubbles, and a redacted decision-log drawer.
Voiceover
A boutique studio does not need more notifications. It needs a system that understands what each client text would change.
[REDACTED_CLIENT] inbound: "Can I get into the evening class?" Intent classified as booking request.
10 seconds
Visual
Capacity, package balance, policy window, and duplicate-booking checks light up one by one.
Voiceover
Becky reads the same state the owner would check: capacity, credits, bookings, policies, and recent thread context.
Decision: full class, active pack, waitlist slot available. Action limited to waitlist offer.
10 seconds
Visual
A drafted reply pauses at a confirmation gate before any charged change.
Voiceover
The moat is not just drafting. It is knowing when not to act, when to ask for YES, and when to escalate.
Guardrail: no credit consumed and no charged late cancel without explicit confirmation.
10 seconds
Visual
An SMS thread shows a concise reply with no owner interruption.
Voiceover
Clients get clear answers while the studio record stays coherent behind the scenes.
Outbound to [REDACTED_CLIENT]: waitlist confirmed, booking held for final confirmation if a mat opens.
10 seconds
Visual
A compact owner summary shows intent, checks, guardrail, and next action.
Voiceover
Every action leaves a readable trail, so the owner can trust what happened without reading a raw transcript.
Owner summary: [REDACTED_CLIENT] asked for class access, accepted waitlist, no payment state changed.
10 seconds
Visual
Three repeated threads resolve into a cleaner dashboard and fewer owner interrupts.
Voiceover
That loop compounds across bookings, packages, late cancels, and payments. Becky becomes the operating layer.
Outcome: resolved client request, preserved policy, logged next action. Video production follow-up: final cut from this storyboard.
BookYourMat