Pricing models explained, August 2026

What an AI voice agentactually costs a brokerage.

Nobody in freight voice AI publishes a price list, so the useful question is not what it costs, it is how you are billed. There are four models in this category and they behave very differently once volume shows up.

Before you compare quotes

The four questions that decide what you actually pay

01

What is the meter?

Minutes, calls, callers, seats, or nothing. The meter decides whether success makes your bill go up or leaves it flat.

02

What counts as a call?

Voicemails, hang-ups, wrong numbers, and transfers. Ask which of those are billable before you sign, not after.

03

Is integration a separate line?

TMS integration is the expensive engineering. Find out whether it is included, one-time, or an ongoing platform fee.

04

What happens in a busy month?

Model your worst week, not your average one. Metered pricing is priced on the average and paid on the peak.

The four billing models

This compares structures, not vendors. Freight voice AI vendors quote privately, so anyone publishing a competitor's exact rate is guessing. Take these questions into your own quote calls.

ModelHow it worksWhere it hurtsBest for
Per minuteYou pay for connected talk time, often rounded upLong calls are the valuable ones. A driver explaining a breakdown is exactly the call you get billed most forVery low volume, or a pilot you intend to keep small
Per callA flat charge per handled call regardless of lengthWrong numbers, hang-ups, and repeat callers can all bill. Define what counts before signingPredictable, low-variance call volume
Per seat or per userPriced like software, by how many of your people it touchesIt has nothing to do with call volume, so you pay for growth in headcount rather than value deliveredPlatforms your team operates rather than a service
Flat monthlyOne price for the workflows and volume band you agreedRequires a real scoping conversation up front rather than a signup formBrokerages that want to automate more calls without the bill moving
Human answering service (hourly or per minute)People reading a script, billed by timeNo system access, so every call becomes a task for the morning, and the worst nights cost the mostMessage-taking only, which is not dispatch coverage

How to build your own number

Step 1: count the calls you actually want automated

Not total call volume. Pull a week and split it: carrier calls on posted loads, check calls, after-hours driver calls, paperwork chasing, appointment scheduling. Most brokerages find that three call types make up the large majority of the volume, and that is what you are buying coverage for.

Where it wins

  • Gives you a like-for-like basis to compare very different quotes
  • Shows immediately whether your problem is inbound, outbound, or after-hours
  • Usually reveals one call type nobody realised was eating a full-time equivalent

What to check first

  • Do not count calls you would never let an agent handle, like claims or accidents
  • Count repeat callers separately, because some pricing models bill each attempt

Verdict: Without this number every quote you get is unanchored and every vendor sounds equally reasonable.

Step 2: price the status quo honestly

The comparison is not against zero. It is against what the phone costs you now: the after-hours coverage you pay for, the loaded hourly cost of the reps taking repetitive calls, and the loads you lose to voicemail. That last one is the biggest and the least measured.

Where it wins

  • Reps are the largest line, and repetitive calls are a measurable fraction of their day
  • After-hours cover, answering service, or an on-call phone all have a real invoice
  • Missed calls have a knowable cost once you know your average margin per load

What to check first

  • Do not inflate the missed-call number to justify a purchase. Use your own margin and a conservative conversion rate
  • Turnover cost is real but hard to defend in a business case, so keep it as a footnote

Verdict: A brokerage that has done this arithmetic negotiates much better, whoever they end up buying from.

Step 3: model the busy month, not the average

Metered pricing looks cheapest in the quote and costs the most in the month you needed it. Take your worst week, multiply it out, and ask each vendor what that month invoices at. The answers get noticeably less comfortable.

Where it wins

  • Exposes rounding rules, minimums, and overage bands that never come up in a demo
  • Tells you whether the vendor has thought about your peak or only about their average customer

What to check first

  • Ask specifically about hold time, transfer time, and failed calls
  • Ask what happens if you double your automation next quarter

Verdict: If a vendor cannot tell you what a peak month costs, you do not have a quote, you have an estimate.

How we price it

Flat monthly, scoped to your call volume band and the workflows you want covered, with the TMS integration work included in the build rather than sold as a separate platform fee. No per-minute meter and no per-seat licensing, because both punish you for the exact behaviour you are trying to encourage. We quote after a scoping call, not from a form.

Where it wins

  • Automating more calls does not increase the bill inside your band
  • A long, messy, valuable call costs the same as a thirty-second one
  • Integration engineering is part of the build, not an upsell
  • One number you can put in a budget and defend

What to check first

  • It requires a real scoping conversation, so there is no instant signup
  • If your volume is tiny and irregular, a metered vendor may genuinely be cheaper and we will say so

Verdict: Built for brokerages that intend to automate the phone properly rather than dabble.

When you should not buy this at all

There are brokerages that should keep their money this year, and it is better to hear that now than after an implementation.

  • You take a handful of calls a day and everybody answers the phone anyway. Automation solves a problem you do not have.
  • Your call volume is genuinely unpredictable and tiny, in which case a metered vendor may cost you less.
  • Your real bottleneck is quoting throughput or order entry, not phone coverage.
  • Your TMS has no integration path and live write-back is the only outcome you care about.
  • Nobody internally will own the escalation rules. An agent with no defined escalation path is worse than voicemail.

Pricing questions brokers ask

Because deployments genuinely differ. A brokerage automating check calls on 200 loads a month and one automating an entire carrier sales desk are buying different amounts of work, and integration depth varies enormously by TMS. That said, a vendor should be able to give you a real number after one scoping call. If it takes three meetings and a procurement process to learn the shape of the pricing, that tells you who they are built for.

Not always. At very low volume it can be the cheapest option and it lets you start without a commitment. It becomes a problem at exactly the point automation starts working, because your bill grows with your success and you end up quietly discouraging the calls you should be encouraging.

That depends entirely on your platform. A documented public API like Alvys is a straightforward build. A system with no public API means EDI work or a structured handoff design, which is more effort for a less complete result. We scope it before quoting, and we tell you when the integration is not worth what you would pay for it.

Cost per handled call at your realistic peak volume, what happens to that number if you double automation, whether integration is included, and what the vendor writes back to your TMS. Those four turn very different pricing structures into one comparable figure.

We scope the build and quote it as part of the engagement rather than as a surprise line item. What we will not do is quote a low monthly and then bill the integration work separately once you are committed.

The honest answer is that it depends on which call type you automate first, which is why we start with the one where your own numbers make the case. After-hours coverage and check calls tend to be the clearest, because both have a measurable current cost. We will not quote you an industry payback figure we cannot substantiate with your data.

Get a real number, not a range

Book 15 minutes with your call mix and your TMS. You will leave with a scoped price and an honest read on whether it is worth doing this year.