What an AI Answering Service Actually Costs in Canada

If you have priced AI answering for a Canadian trades business, you have probably noticed the same thing we did: nobody quotes the number you actually pay. The headline is a per-minute rate, and the bill is something else.

This is a plain breakdown of how the pricing works, what stacks on top, and how to work out your own cost before you commit to anything.

The Three Pricing Models

Flat monthly plans. One line item, a usage credit, and one published rate for everything you use. You can predict next month's bill before it arrives. This is how our own plans work.

Usage-based platforms. A per-minute platform fee that covers orchestration and nothing else. You bring your own language model, your own speech-to-text and text-to-speech, and your own telephony carrier, and each of those bills you separately. Vapi, Retell and Bland sit here.

Human answering services. Quoted per call or per minute, usually against a monthly retainer and a contract. Pricing is rarely published, so the only way to compare is to ask for a quote and read the minimum commitment.

What the Headline Rate Leaves Out

On a usage-based platform, the advertised figure is the platform's cut. Four things are billed on top:

• The language model — Anywhere from about a cent a minute for a small fast model to twelve for a frontier one
• Speech-to-text and text-to-speech — Half a cent to nine cents a minute, depending on voice quality
• Telephony — Under a cent to five cents a minute, per leg, through your own carrier
• Concurrency — Roughly $10 a month for each simultaneous line beyond the included set

Add them up and a $0.05/min platform fee becomes somewhere between $0.07 and $0.31 a minute. That is not a criticism of the model — it is the right shape if you want to swap components and tune cost. It is just not a number you can budget from.

There is a second cost that never appears on any invoice: the work of assembling it. Somebody has to pick the models, hold the carrier account, and own it when a voice provider has an outage mid-call.

What Voismatic Costs

Two plans, in Canadian dollars, billed in CAD so there is no foreign transaction fee on your card:

Both include a phone number, unlimited knowledge sources, and unlimited agents. The monthly price is your usage credit: voice and SMS draw from the same balance at the plan's rates. Neither plan has a hard cutoff — once the credit is spent, usage keeps billing at the plan rate rather than silently dropping calls, because a call that gets refused because you hit a cap is the exact call you bought the service for.

The rate column is the one to compare against a per-minute quote. It is the monthly price divided by the minutes that credit buys, and it is the honest comparison: a plan with a larger credit buys a lower rate, and the rate does not move as you use more.

Working Out Your Own Number

Three inputs decide everything:

1. How many minutes a month. Count the calls you currently miss or send to voicemail, not your total call volume. Multiply by your average call length — for trades intake, two to four minutes is typical.

1. How many concurrent calls. One line is enough for after-hours. Storm season, a furnace failure in a cold snap, or any event that puts every crew in the city on the phone at once is where concurrency starts to matter.

1. What a captured job is worth. This is the number that decides whether any of it is worth doing.

Our voice AI cost calculator runs the first two against published rates from every major platform, and the missed call calculator runs the third.

The Break-Even Is Lower Than the Debate

Here is the part that usually ends the pricing conversation early. At $250 a job, a single captured job covers a Starter plan for six months. At $99/mo on Pro, one job every other month covers it.

Two missed calls a day is roughly 25 lost jobs a year, assuming half of missed callers never call back and one in ten quotes closes. At $250 a job that is $6,250 walking out the door, against $1,188 a year for the plan that catches it.

That ratio is why we publish a flat price instead of a per-minute one. If the math only works at scale, it is not a product for a three-truck shop.

What to Ask Before You Sign

• What is billed on top of the advertised rate? Model, speech, telephony, concurrency, and any compliance add-on
• What happens when I exceed the included usage? A published rate, or a hard cutoff
• Is the phone number included, or a separate carrier account?
• What currency am I billed in? A USD invoice on a Canadian card adds roughly 2.5% in foreign transaction fees
• What is the minimum commitment? Month-to-month, or an annual contract

The Honest Summary

If you want the cheapest possible per-minute rate and you have somebody who enjoys wiring up model providers, a usage-based platform will beat a flat plan on raw cost. If you want one bill you can predict and somebody else owning the stack, a flat plan is the cheaper answer once you count the assembly.

For most trades businesses, the question is not which platform is cheapest per minute. It is whether the phone gets answered at 2 a.m. — because the cost of the answer, at any of these rates, is smaller than the cost of the job you lose.

Competitor figures above are published rates as of July 2026, converted from USD at 1.37. See the full pricing comparison for the per-provider breakdown and sources.