Metilda Clinical Insights
AI Receptionist Setup Time: What a Cliniko Clinic Actually Needs Before Go-Live
A generic AI receptionist can reportedly go live in five minutes flat, according to vendor marketing published this year. A Cliniko allied health clinic is a different kind of setup problem. Before an AI receptionist can safely quote a bookable slot, someone has to get three settings right for every practitioner, every appointment type, and every clinic location: the same capacity limits, minimum-notice windows, and booking-window rules that Cliniko's own scheduling API enforces. A clinic with three practitioners, four appointment types and two locations isn't one setup task. It's roughly two dozen distinct combinations that each need to be configured correctly before the AI receptionist can be trusted not to offer a slot that doesn't actually exist. That's the real AI receptionist setup time question: not how fast the phone number goes live, but how long until every one of those combinations is right.
The "Five-Minute Setup" Myth in AI Receptionist Marketing
Search for AI receptionist setup time and the marketing is consistent: connect a number, write a greeting, go live. One vendor guide published 22 September 2026 breaks it down almost to the minute: 30 minutes for account and phone configuration, 45 minutes for a custom greeting, then "answering calls within the first 5 minutes" of day one, with calendar and CRM integration folded into a 2.5-hour "intelligence phase" on day two. A separate healthcare-adjacent guide, published a month earlier, makes a similar pitch while being more candid about why: "the variable is your configuration, not the technology," it admits, before noting that "complex routing, multiple locations, and many custom call types add configuration time," without saying how much.
Neither guide, notably, is written for a clinic running Cliniko with more than one practitioner. A single-location business answering one undifferentiated queue of calls really can plug in a phone number and go live inside a day. A multi-practitioner allied health clinic is booking against a scheduling system where availability is practitioner-specific, appointment-type-specific and location-specific all at once. None of that shows up in a 75-minute onboarding script built for a generic front desk.
What Actually Drives AI Receptionist Setup Time at a Cliniko Clinic
Cliniko's own developer documentation is the clearest evidence of what's actually involved. The Available Time endpoint, the one any AI receptionist has to query to find out whether a slot is genuinely bookable, requires a Business ID, a Practitioner ID, and an Appointment Type ID just to run. That's three dimensions of configuration before a single date is even considered.
The Six Cliniko Settings That Decide What "Available" Means
Cliniko's documentation lists six configuration settings that govern what that endpoint returns for any given business/practitioner/appointment-type combination:
- Appointment capacity limits — how many bookings of that type are allowed per day segment (1–5, or unlimited).
- Minimum advance notice — anywhere from "Now" to four weeks ahead.
- Booking window — how far into the future availability is shown, from one day to two years.
- Business enablement — whether that location is enabled for online/API booking at all.
- Appointment type enablement — whether that specific type is bookable through the API.
- Practitioner enablement — whether that practitioner is included in API availability at all.
Get any one of these wrong for a single combination, and the AI receptionist either offers a slot the practice never intended to release, or silently refuses to offer a slot that's genuinely free. Neither failure mode is visible from the phone call itself — both only show up later, as a frustrated patient or an empty diary.
A Worked Example: Counting Your Clinic's Real Configuration Load
Take a mid-sized allied health clinic: 3 practitioners, 4 appointment types (initial consult, standard follow-up, telehealth, home visit), 2 business locations. Assume every practitioner offers every appointment type at both locations. Real clinics rarely match that exactly, but it's the clean starting number.
3 practitioners × 4 appointment types × 2 locations = 24 distinct availability combinations.
Each of those 24 combinations touches up to six of the settings above — so the realistic configuration surface for that single, fairly ordinary clinic is closer to 144 individual checks, not one toggle.
A clinic doesn't have one setup task — it has as many as practitioners × appointment types × locations, each gated by up to six Cliniko-side settings, which is the real source of AI receptionist setup time at any clinic bigger than a solo practice.
A solo practitioner running one appointment type out of one location has a configuration surface of exactly 1, which is probably why the five-minute pitch holds up fine at the smallest end of the market and stops applying almost immediately past it. Double the practitioner count or add a second location, and the number doesn't grow by addition. It multiplies.
None of the Three Vendor Sources Checked Will Give You a Number
Before writing this, we checked what three real sources actually say about setup time for a multi-practitioner context, specifically to see whether any of them would commit to a figure.
| Source | What it claims | Concrete multi-practitioner timeframe given? |
|---|---|---|
| Solium AI's own Cliniko listing | "Fully managed onboarding and optimisation, meaning no setup effort for your team" | No |
| getnextphone.com's setup timeline guide | Live in "5 minutes," full day-one build in roughly 75 minutes | No (built for a single generic queue, not multi-practitioner scheduling) |
| Sonant's setup-time piece | "The variable is your configuration, not the technology" | No — names the variable, doesn't quantify it |
Taken together, that's a pattern worth naming plainly: the category is comfortable describing the complexity and uncomfortable quantifying it, probably because the honest answer ("it depends on your practitioner count, appointment types and locations") doesn't fit neatly on a pricing page. Treat the absence of a number as a prompt to ask any vendor, Metilda included, for one specific to your own clinic, not as evidence the number is small.
A Realistic AI Receptionist Setup Timeline, Stage by Stage
Stripped of marketing, the stages that actually have to happen, in order, look like this:
Stages A and B genuinely can take hours, matching the generic marketing. Stage C is where the timeline actually lives, because it's the only stage whose length depends on the clinic's own size rather than the vendor's software.
This is also where Metilda's own positioning is specific rather than generic. Because it's built directly against Cliniko's calendar and appointment-type structure, instead of a universal scheduling sync that has to work identically for a hairdresser, a plumber and a physiotherapist, the configuration happening in stage C is being set against the real Cliniko settings described above. Stage D doesn't have to stay a black box either: Metilda's client portal lets a clinic review the actual call transcripts from that testing stage directly, rather than take a vendor's word that the test calls went well.
Before You Sign: Three Questions About Setup Time Worth Asking Any Vendor
Skip the demo-day stopwatch. Ask these three questions instead, every one answerable in writing, before a contract, by any vendor telling the truth about their own product:
- "Walk me through exactly which of my practitioners, appointment types and locations will be enabled on day one, not eventually." A vague "everything gets set up" answer means the vendor hasn't actually looked at your Cliniko account yet.
- "What happens to a patient who calls before every combination is configured?" A clinic with 24 combinations rarely gets all 24 right simultaneously on day one, so the honest answer involves a phased rollout, not a single flip of a switch.
- "Can I see the actual settings you've configured for each practitioner and appointment type: capacity limits, minimum notice, booking window?" If the vendor can't produce that list, they're asking you to trust a black box for a decision that directly affects whether real patients get booked correctly.
FAQ
How long does AI receptionist setup actually take for a Cliniko clinic?
For a solo practitioner with one appointment type and one location, hours is realistic — the configuration surface is a single combination. For a clinic with multiple practitioners, appointment types or locations, the honest answer scales with that count: a 3-practitioner, 4-type, 2-location clinic has 24 combinations to configure correctly, each gated by up to six Cliniko settings, which is real work regardless of how fast the phone number itself connects.
Why do some AI receptionist vendors advertise a five-minute setup?
For the simplest case, one queue, one calendar, no practitioner-specific or location-specific availability rules, that claim can be accurate. It's built for businesses where every incoming call is interchangeable, which doesn't describe a multi-practitioner clinic booking against practitioner-specific Cliniko availability.
Does setup time really scale by multiplication, not addition?
Yes, based on how Cliniko's own Available Time API is structured: it requires a Business ID, Practitioner ID and Appointment Type ID together, so each additional practitioner, appointment type or location adds a new combination against the existing ones rather than one extra item on a list. Three practitioners and four appointment types isn't seven things to configure — it's twelve combinations, before locations are even counted.
What should I actually check before trusting an AI receptionist to quote real appointment slots?
Ask the vendor to produce, in writing, the specific capacity limits, minimum-notice windows and booking windows configured for each of your practitioners and appointment types — not a general assurance that "it's all handled." If they can show you the settings, you can verify them against what your practice actually intends. If they can't, you're trusting a black box with real patient bookings.
Does a longer setup mean a worse product?
No. If anything, the opposite can be true. A vendor that spends real time mapping every practitioner, appointment type and location combination before go-live is doing the configuration work that keeps an AI receptionist from confidently offering slots that don't exist. What should concern you is a vendor who can't explain what that time is actually being spent on, whatever the clock says.