Most salon software sends texts. Ada holds the conversation. She answers the question, proposes two real open times, and books the appointment before the client shops somewhere else.
About the time it takes a client to reread what they just sent.
Question, price, two real times, confirmed. Not a callback tomorrow.
One client or forty, 2 PM Wednesday or 11 PM Sunday. It doesn't queue.
Unedited screen recordings from real conversations. Watch the second one especially: the client never says a service name, and she still ends up on the right stylist's calendar.
“Was recommended Isabella, when can she give me a haircut tomorrow?” Ada takes it from there.
Bubbly, calm, quiet, talkative. Ada asks the question no booking form knows how to ask.
“Since it's past the 8-week window, are either rows slipping, lifting, tangled, matted, uncomfortable, or really grown out?”
“Was recommended Isabella, when can she give me a haircut tomorrow?” Ada asks hair length and a name in a single short message, then goes to the live calendar for times. That number had no client record a minute earlier.
The client wants something brighter that grows out naturally. That is an outcome, not a service. Ada translates it into balayage, picks the specialist for it, and moves to booking. This exact moment is where online booking loses people who don't know your menu.
A client asks for a basic extension move-up. Ada checks the last appointment, sees it lands past the eight week window, and asks whether rows are slipping or grown out before she books anything. One row is slipping, so she books the longer service instead.
The tell is never the answer, it's the phrasing. Ada is built with a specific voice and a list of phrases she is forbidden from using, because one corporate sentence undoes an entire thread.
Em dashes are on that list too. Almost nobody types one in a text message, and it has quietly become the clearest signature that a machine wrote something.
Ada introduces herself as your AI concierge in her first message of a thread, and never again after that, because re-introducing yourself is its own kind of robotic.
When someone asks point blank whether they're talking to a person, she confirms it with confidence and pivots straight back to the calendar. She never calls herself a bot, a chatbot, a virtual assistant, or a system, and if a client uses one of those words, she corrects it gently. Concierge is the framing, because that's what she actually is.
“How much?” is where most salons lose the booking. Quote it flatly and the thread dies. Hedge and you sound ashamed of your own prices. Ada has four hard rules and she does not break them.
If the client didn't ask about cost, Ada doesn't raise it. Volunteering price invents an objection nobody had.
No “it's a bit pricey, but…”. Confidence in your pricing is itself a sales tool, and hedging reads as shame.
Asked directly, she gives the number. Dancing around it is the fastest way to lose trust in a text thread.
Not to close, not to calm someone down. The price is the price, and the reply ends with two open times instead.
“$185 for highlights, want me to find you a time?”
That single move handles most price questions. When someone pushes back after the number, she doesn't defend it. She asks what they've been paying and what they've been getting, then proposes a time around the answer.
Anyone can list open slots. Choosing the right two is the difference between a booking and a thread that goes quiet.
Same-day is the strongest close there is, and tomorrow is second. She reaches for those first, every time.
Same-day offers sit at least an hour out. “Can you come in right now?” reads as either pity availability or a machine guessing.
One morning and one afternoon. Stacking 10 AM and 11 AM wastes the second option, and a wall of slots makes people close the thread.
“Schedule's pretty tight this week” is a sales line and Ada is allowed to use it. Claiming a slot is open when it isn't is a double-booking, so every time she names comes from a live look at your calendar, and availability is re-checked every single time she's about to offer or confirm one, no matter what she said five minutes ago.
Ada never pushes harder. Hesitation is information, so every stall has a matching question designed to surface what's actually in the way, then she reflects it back and proposes a time.
“Let me think about it.”
What does the perfect salon and stylist look like for you?
“My last place messed it up.”
What didn't you love about your last appointment?
“That's more than I usually pay.”
Can I ask what you've been paying and what you've been getting?
“Just looking for now.”
Of course, what's prompting the change?
“It's impossible to style at home.”
How much time do you usually spend styling in the morning?
“I'm comparing a few places.”
Totally fair. What are you comparing us against?
“After my trip.” “Maybe next month.” Ada doesn't push and she doesn't drop the thread. She schedules herself a follow-up against that client's record for roughly when they'll be ready, and saves the reason in their own words: traveling until mid-July, waiting until after the wedding. That note is the only context she has when she reaches back out, and it's what makes the follow-up land instead of reading like a blast.
See how follow-ups work →Ada will not cancel an appointment the moment someone asks. She offers to move it, casually and only once, and only cancels if the client makes it clear they don't want to rebook.
That one reflex is the difference between a hole in Friday and a client on Tuesday. When a cancellation does go through, the slot doesn't just sit there: anyone who wanted in sooner but had to book further out hears about the opening first.
Anything that would normally push a client to a login screen, Ada sends as a link they tap once, then she finishes the booking in the same conversation.
Require a saved card or a deposit before a booking is confirmed. Ada texts a no-login link and completes the appointment once it's done.
Consent forms, inspiration shots, current-hair photos. She sends the link and keeps reminding until it's filled in.
Reminders, day-of logistics, and the check-in text that tells the stylist their client just walked in.
Six steps between an inbound message and a booked appointment. None of them involve guessing.
Before Ada writes anything, the text is scored for whether it deserves a reply at all. “Thanks” and “ok” get silence, because answering those is how a client works out they're talking to a machine.
Unknown numbers get a client record created automatically, so there is no anonymous path and nothing falls outside your database. The moment Ada learns their name, she writes it to that record.
Ada doesn't hold standing access to your data. Each conversation mints a token scoped to that one client, and the backend re-checks ownership on every appointment she touches.
Services and prices stay good for a conversation. Availability never does. It goes stale in minutes as other people book, so she re-checks before every single time she offers or confirms.
Service, provider, and a specific time the client said yes to. A date alone is not a time: “Monday” gets two options and a wait, never a silent pick.
Each lookup and each change is logged with what it did and when. If you ever want to know why a client ended up on Thursday at 11, the answer is on the record.
The reason you can leave her on overnight is that the list of things she can't do is as deliberate as the list of things she can.
If a lookup fails, she says her system is being slow and tries again. If it fails twice, she points the client at your team. She will never fake a confirmation to cover an error.
How Ada handles your clients over SMS.