JaneApp manages your bookings. The calls that cost you money happen before there's anything to book.

01 What those calls actually cost

Ask a Winnipeg clinic where the phone time goes and it's rarely the booking itself. It's the questions before the booking: "my lower back's been sore two weeks — should I see a physio or a massage therapist?" and "how soon can I get in?"

Now put numbers on it:

• Every one of those calls needs a person to read the symptom, match it to the right therapist, and check the schedule — a medical receptionist in Canada runs roughly $16–$29 an hour, and a full-time front desk is a $38,000+ a year line item before benefits.
• If your therapists are the ones answering, the cost is worse: every hour on the phone is a treatment hour not billed.
• If nobody answers fast enough, industry data is blunt — most callers won't wait on hold past about a minute, and a large share hang up before they ever reach a person. You don't lose one appointment; you lose the whole client relationship that appointment would have started.

02 Where JaneApp stops

JaneApp is built for Canadian clinics and it's very good at what it's built for — holding the slot, running the bill, sending the reminder. What it doesn't do is judge: it can't hear "stiff neck since a car accident" and decide that's a physio intake, not a relaxation massage.

03 One wrong booking can cost you the whole invoice

Here's the version of this problem that hits the bank account directly. Under Manitoba Public Insurance rules, massage therapy for a collision injury is only covered when it's delivered under a physician, chiropractor, nurse practitioner, physiotherapist, or athletic therapist. A new patient calls after a fender-bender, whoever picks up books them straight into a relaxation massage — and that claim can come back denied. Now you're either chasing the patient for the bill or writing it off.

A similar risk shows up on WCB claims and on extended-health plans, where coverage rules and direct-billing arrangements vary by insurer and aren't something a front-desk person can be expected to just know from memory. This is exactly the kind of judgment call a generic offshore FAQ bot can't carry: it doesn't know how direct-billing actually settles in Manitoba, which insurers pay this clinic directly, or what a patient can expect back. Answer that wrong once, on a first call that was supposed to build trust, and it costs more than the call did.

04 Why "just get an AI phone answering service" doesn't fix this

You've probably seen ads for AI phone receptionists that answer instantly, 24/7, and book straight into your calendar. They're real, they're cheap, and for a restaurant or a salon they work fine.

For a clinic that takes MPI and WCB claims, that's not a smaller version of the solution — it's the exact failure mode in section 03, just running faster and around the clock. A generic answering service doesn't know that massage therapy for a collision injury only counts when it runs through a physician, chiropractor, nurse practitioner, physiotherapist, or athletic therapist. It hears "my back hurts," checks the calendar, and books the first open slot — the same wrong booking, just automated and unsupervised. Most of these tools also process the call on servers outside Canada, which matters more for a health complaint than it does for a dinner reservation.

Full automation isn't the missing piece here. The missing piece is judgment applied before anything gets booked, with a person still making the final call on anything that touches a patient's health information or an insurance claim.

05 The deeper problem: all of this lives in one person's head

Even when your front desk handles every one of these calls perfectly, the way it works today is fragile in three ways:

• The judgment isn't transferable. A front desk person with five years in knows that car-accident callers need an assessment first, that prenatal goes to one specific therapist, which insurers this clinic direct-bills. When she leaves, all of it walks out the door with her — and front-desk and medical-assistant roles in health practices commonly turn over at 25–40% a year, which means owners are re-hiring and re-training every couple of years while the knowledge keeps draining away.
• A person clocks out. After hours, over lunch, while checking in the patient standing at the desk — the phone rings into nothing. The inquiry that came in at 7pm is booked somewhere else by 9am.
• Quality collapses under load. With three things happening at once, even a great front desk misses the one question that matters — "was this from an accident?" — and that one missed question can be a denied claim.

None of this is a criticism of your staff. It's the shape of the job: high-judgment work, running on memory, under interruption, with turnover built in.

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

I build the triage layer that sits in front of JaneApp — not a replacement for it, and not a full-automation answering service either. Every new inquiry runs through four steps:

1. Intake. The patient's message or call is captured — complaint, how long it's been going on, what they're hoping for.

2. Insurance routing. The system reads for what changes everything — not just the words "car accident" or "WCB," but the way people actually say it: "someone rear-ended me last week," "this happened on a job site," "insurance is handling it." Those inquiries get flagged and routed toward an assessment-first booking — so a claim-killing "relaxation massage for a collision injury" never gets suggested in the first place.

3. Therapist matching. The symptoms are checked against your team's actual specialties — prenatal, lymphatic drainage, TMJ, sports rehab — so the suggestion is the right practitioner, not just whoever has an opening.

4. Sourced answers. Direct-billing and coverage questions get answered from your clinic's own documented rules, with the source attached. If the answer isn't documented, it says "not found" instead of guessing.

The output of all four is a drafted suggestion — recommended service, recommended therapist, realistic wait, intake note — delivered to your front desk. None of this touches the clinical chart: the AI only sees what a caller volunteers before there's a patient file to speak of, drafts a suggestion, and hands it to a person. The treatment record stays exactly where it's always been — in Jane.

And nothing books itself. A person on your team reviews the match and confirms it before a patient is placed. That's a deliberate design choice, not a workaround: when a suggestion touches a car-accident or workplace-injury claim, or a new patient's health information, a human making the final call is the right shape — regardless of what any platform's API does or doesn't allow.