Your booking app fills the slot. It can't tell whether this client should be in it.

01 What the booking-before-the-booking actually costs

Ask a Winnipeg PMU studio or med spa owner where the money leaks and it's rarely the treatment itself. It's the first contact — a DM or a call, usually after hours, from someone comparing two or three studios before she books. Miss it and she books with whoever answered first.

Put rough numbers on it (US industry figures, directional — no Winnipeg-specific study): a med spa visit runs $500–700, busy studios miss as much as a third of their calls while they're with clients, and at roughly a 30% inquiry-to-booking rate even three missed calls a day adds up to six figures a year in lost bookings. No-shows run around 20% on top of that. But the expensive part of that first contact isn't only the missed call — it's the wrong booking. The judgment on a med spa's first message isn't "which slot is open." It's "can this person have this service at all, with this technician, and on what deposit terms."

02 Where your booking software stops

Whatever you run — Vagaro, Boulevard, Mangomint, GlossGenius, Square — has added real AI in the last year, and it's good. Boulevard's chat assistant, Billie, answers texts and web chat 24/7 and books straight into the calendar; a separate AI Workforce line handles the phone the same way, and Smart Scribe writes chart notes from the consult. Vagaro plugs into third-party AI receptionists that book straight into your calendar. Mangomint's Connect feature consolidates calls, texts, and web chat into one inbox. GlossGenius offers custom intake forms clients fill out before they arrive.

Look at where all of that lives: confirming a booking, sending a form, writing the chart afterward, marketing to fill gaps. The moment that actually carries risk — is this client's requested service safe for her right now — happens on the first message, before any of it. The intake form that would catch it gets filled at the chair, after the deposit, after the drive in.

03 One wrong booking is a contraindication incident, or a deposit dispute

Here's the version that hits the chair. A client books lip blush and she's newly pregnant and didn't think to mention it — pigment retention and numbing agents make it a no-go, and you find out at the consult. A brow client is four months off Accutane, when the standard wait is a full year. A client on daily aspirin books microblading, and the bleeding thins the pigment and wrecks the retention. Each is a real contraindication — pregnancy or nursing, Accutane within the year, blood thinners, keloid-prone skin, an active skin condition, recent filler or Botox — and each is a question that should have been asked before the deposit, not discovered at the chair.

To be clear about what this is and isn't: this isn't medical advice, and the system doesn't make the medical call. These are the questions your own intake and consent forms already ask — surfaced at booking instead of at the chair. The final judgment always stays with the technician.

The deposit is the other half of it. When a client is turned away for a contraindication after she's paid, you're arguing your policy on a first interaction that was supposed to build trust. Getting the screen right before the deposit is cheaper than getting the refund wrong after.

04 "Can't I just get an AI receptionist that screens for contraindications?"

Fair question — and this time the honest answer is yes, they exist, and some are decent. Generic AI voice agents (Retell paired with GoHighLevel, CallSetter, and similar) already advertise contraindication screening: they collect health history and medications and escalate flagged calls to a medical director before booking. Some vendors in this space quote live deployment inside 48 hours. A build runs a few thousand dollars plus per-minute voice costs. If you're a high-volume med spa that just needs the phone answered and the obvious flags caught fast, one of these might be enough — and I'll tell you so.

But for a PMU or med spa specifically, they share three real gaps:

• The screen is generic, not built around your studio. It knows "pregnancy" and "Accutane" as keywords. It doesn't know that your microblading artist and your laser tech hold different qualifications, that a specific pigment line carries its own restrictions, or exactly which services your studio will and won't do on a given flag. That's config you'd maintain, not a system built around your actual menu and your artists.
• The default is full automation on the clear path. These agents escalate the calls they flag — but the calls they judge "clear" get booked automatically, with nobody at your studio approving that booking. For a service where a missed question is a contraindication incident, "only the escalations get a human" is the wrong default. Mine flips it: a person confirms every booking that touches a health flag or a deposit, not just the ones the AI wasn't sure about.
• The call audio usually leaves the country. A prospective client describing her pregnancy, her medications, her skin condition is sharing health information, and most of these tools process the call on US servers under their own retention terms. Worth asking any vendor exactly where the audio lives, and for how long.

On speed: a generic agent can genuinely be live faster than a custom build — that's a real trade-off, not a claim I'll talk you out of. What you're trading for that speed is a screen built for the average studio, running on someone else's servers, with everyone not flagged getting auto-booked. If your studio's flags, artists, and policies are simple enough that the generic version covers you, that speed advantage is worth taking. If a wrong booking is expensive enough at your studio that you want every booking — not just the flagged ones — reviewed by a person before it's confirmed, that's the trade the extra build time buys you.

05 Why the booking platforms won't build this part

It's reasonable to wonder whether Vagaro or Boulevard will just add this next quarter. Look at where their AI investment goes: booking, reminders, marketing, charting — features where one model works identically for a nail salon, a blow-dry bar, a massage studio, and a med spa. That's what a horizontal platform scales.

Per-vertical contraindication logic is the opposite. It's specific to PMU and injectables, it changes by service and by technician qualification, and when it's wrong there's real liability attached. That doesn't scale across their whole beauty-and-wellness base, so they leave it to third-party configuration instead of building it in. Which is why the right version of this gets built around your studio, not toggled on inside a platform.

(This is my read of where these platforms are pointing their AI, from their published features — not an official statement that they'll never build it. What's verifiable is that none of them ship pre-booking contraindication triage today, and all leave phone screening to third parties.)

06 What I build: four checks on every new-client message, before anything is booked

I build the screening layer that sits in front of whatever booking software you already run — not a replacement for it, and not a full-automation answering service either. Every new inquiry runs through four steps:

1. Intake. What service the client wants, whether she's had it before, and the skin and health background the service actually requires — captured from the call or message in her own words.

2. Contraindication routing. The system reads for what changes everything — not just the words "pregnant" or "Accutane," but the way people say it: "we're expecting," "I just finished my acne meds," "I'm on a baby aspirin." Anything that hits a flag — pregnancy or nursing, Accutane within the year, blood thinners, keloid-prone skin, an active skin condition, recent filler or Botox — is routed to a consultation and patch test first, never straight into a full service. These are the questions your own consent form already asks, surfaced before the deposit.

3. Service and technician matching. The request is checked against your team's actual qualifications — microblading, machine PMU, microneedling, and laser are different certifications — so the suggestion is the right artist for the service, not just whoever has an opening.

4. Sourced answers. Deposit rules, whether a touch-up counts as a new booking or a follow-up, the reschedule policy — answered only from your studio's own documented policies, with the source attached. If it isn't documented, it says "not found" instead of inventing an answer.

The output of all four is a drafted suggestion — recommended service, recommended artist, any flags, and a realistic next step (book, or consult first) — handed to your front desk.

And nothing books itself. A person on your team reviews the suggestion and confirms it before a client is placed — and that's a deliberate design choice, not a limitation. When a suggestion touches a contraindication flag, a health disclosure, or a deposit, a human making the final call is the right shape. Your studio approving a booking means your studio owns it, with a traceable record of which words and which policy the suggestion came from — not a black box that booked on its own.

Three things that matter for a Winnipeg studio, and that a generic subscription agent isn't set up to offer:

• Your client data stays in Canada — hosted on AWS ca-central-1, call recordings kept or deleted on your studio's terms, not a US vendor's.
• You own the finished system — built around your menu, your artists, and your policies, and yours to keep, not rented month to month.
• I'm in Winnipeg — when your services, flags, or deposit rules change, you're talking to the person who built it, and I stay through your front desk actually using it, not just a handoff.