What this doesn't do.

Where the arithmetic stops working, and how you would know

Below roughly 100 inbound calls a month, the full service usually doesn't pay for itself. At 40 calls there is little to recover and the monthly is the same. That is arithmetic on our price, not a measurement of your line — we have not counted your evenings and neither has anyone else. The 30-day pilot is what counts them: your real calls, on your real number, for $1,000. If the count comes back saying it isn't worth it, we will say so and stop there. A cheap answering app or a part-time person may well serve you better, and that is your decision to make with a number in front of you rather than ours to make for you.

Same if the phone isn't your problem. You book everything that comes in and the issue is lead volume, marketing spend or pricing: we are the wrong purchase. The diagnostic is partly built to find that out — and then you paid $2,500 to avoid paying us for a year to fix the wrong thing.

If whoever answers your phone after hours can already put the job on your schedule, you have what we sell. We will say so on the call and stop there.

What it says about money, and what it won't

Give us your service-call fee and the assistant says it before it confirms the booking — not when the caller thinks to ask — with whether you waive it on the repair. A customer never told what the visit costs is a cancelled appointment and an argument you lose. Give us no number and it says nothing about price at all. It will never guess one, and it will never tell someone the visit is free.

It will not put a firm price on work nobody has looked at. There is nowhere yet to give it your price ranges, so if a caller insists on a figure for the job, the ballpark it gives is its own, not yours.

Callbacks go to the number they called from

The assistant books against the caller ID your carrier reports, never a number read out loud: a mis-heard digit is a reminder sent to a stranger. Asked to use another number, it takes the note, says so, and keeps the one they called from. A withheld number it asks for and reads back digit by digit.

Human transfer is built, and unproven

Transfer is off until you give us a number. With one set, a caller asking for a person hears “let me get someone on the line” and is connected; if nobody picks up the assistant comes back, takes details and marks the callback urgent. With no number — how it ships — it is told it cannot transfer, so it takes details instead of promising what it can't do.

We have not run a transfer on a live customer call. Built for it, unproven at it — a thing we configure and test with you at installation, not one to buy on. After hours it is mostly moot anyway: there is nobody to transfer to.

We text you. We do not text your customers yet

To you, today: a message the moment a job is booked, and your morning and end-of-day reports — by email, or by text if you would rather have them that way.

To your customers, not yet. A text to a homeowner goes out as your business, and US carriers require that business to be registered with them before it can send. That registration is yours to file and theirs to approve — we hand you everything the form asks for, and we will not run a single send on your customers before it clears. Confirmations, appointment reminders, review requests and reactivation campaigns all sit behind it, and the fines for getting it wrong land on you, which is why we would rather wait than be helpful about it.

The one message we would be entitled to send — a text back to someone who just rang you — is built and also not live. Replying to a caller is a different consent position from messaging a list; that argument holds, calling it buyable does not. Every send on that path is refused before it leaves us and we have never watched one arrive. Neither is for sale, and the reply is the one that comes back first.

Invoicing and payments aren't available yet

Taking your customers' card details means handling money that isn't ours, and that path is being rebuilt. Until it is done, Full Operations is not open.

You can subscribe to the schedule. You cannot sync with it.

Jobs the assistant books go onto your schedule in the client portal, checked against your hours and your technicians. A private link subscribes Google Calendar, Outlook or an iPhone to them.

It only reads. Nothing in your calendar comes back to us. A job your office books directly — a walk-in, a call to somebody's cell, a repeat customer texting the owner — the assistant cannot see it, and will book on top of it. True only while bookings come from one place. With a dispatcher, be careful.

No field-service integration: not Housecall Pro, not Jobber, nothing that writes back. If your dispatch lives in one, somebody copies jobs across. On the roadmap, and a fair reason to wait.

We have not run this feed on a contractor's calendar for a full week. Built for it, unproven at it — we set it up and watch it with you at installation.

Two locations are two setups, and there is no combined view

We have not run a two-location installation. Each location gets its own line, its own assistant and its own Monday scorecard, set up exactly the way a single shop is — and there is nothing today that shows you both locations in one place, or reports across them. If that is what you want, this is not it yet.

You are not turned away for it. Being first is priced rather than refused: both lines on the thirty-day pilot for $1,500 instead of $2,000, for the first two shops, until 31 October 2026. After that each location is priced the same as a single shop, because it is the same work at each.

Things it gets wrong

• Heavy accents and bad connections. Better than it was, worse than a person. On a poor line it asks people to repeat themselves.
• Callers who won't stay on topic. Good at interruptions, less good at a long story with the address buried in it.
• Quoting the job. No firm price on a job nobody has looked at, and no ranges of yours to quote from yet — pressed, it offers a rough ballpark of its own. Your service-call fee is handled above.
• Anything genuinely unusual. A situation your scripts never anticipated gets a callback, not a confident wrong answer. Correct, and still a limitation.

We are a small company

Support is business hours: 4 hours to a response, 1 hour if your call flow is down. No 24/7 desk. No published uptime figure either — telephony, model and hosting providers we do not control decide it, and a number we cannot enforce would be the first dishonest thing on this site.