What will AI usage cost me each month on top of your fee?

Why this is now the biggest unanswered question

Nine of the 58 sales calls in our backlog since June asked some version of this, and it is the fastest-rising question we hear from clinics, coaches and everyone else. It rarely arrives as a clean question. It arrives as "what does credits mean, is that tokens?", "how do you replenish them?", "is the video tool expensive?", or, from clients already running, "I need my credits." The honest reason it keeps coming up is that nobody has shown owners the three bills side by side, so the third one feels like a surprise.

Here are the three. Our fee pays for the build and the install. Ad spend goes straight to Meta or Google and never passes through us. AI usage is the third bucket: the subscriptions and credits that pay the companies whose models your roles run on. We don't mark it up and we don't carry it, because we want you to see it, own it, and be able to turn it down.

What a credit actually is

Think of it like a utility meter. A text model is billed on tokens, which are chunks of words going in and coming out. A plan gives you a monthly allowance of that; heavier models burn the allowance faster, lighter models slower. Image and video tools work on credits instead: a set number renew each month, each generation spends some, and in most tools what you don't use does not roll over. That last part matters. On a recent call a client had been paying for a plan and never opening it, so we started treating credits the way a strength coach treats a gym membership: use it or lose it, so plan the sessions.

That is also why we route image and video generation through one assistant rather than logging into five tools. The video platform we use is a hub for several models, and your role reaches it from the same place it does everything else, so you are not learning five billing pages or guessing which one is draining.

What drives the number up or down

Three things, in order. First, how many roles run every day. A role that drafts a morning brief costs almost nothing; a role that rewrites every lead conversation in real time costs more. Second, the model tier. Every provider has a fast, cheap model and a slow, expensive one, and owners often ask to "flip it to the big one until it's done." That is fine for a build week and a waste for a daily routine, so each role gets the lightest model that does the job and only the hard tasks get the heavy one. Third, images and video. These are the most expensive things you can generate per click, and the easiest to cap: decide how many ads a month you need and buy the plan that covers that, not the one that covers a studio.

How we keep it bounded

Before launch we write down which roles run on what schedule, which model tier each uses, and the monthly cap for usage, the way we write plan B for the ads before they go live. Most clinics start on the lowest plan that covers the build and stay there; we move you up only when a specific role is hitting the ceiling, and we tell you which one. Then a scheduled mid-month check reports how much of the allowance is gone so a plan never runs dry on the 20th and nothing gets quietly upgraded. If the number ever matters more than the output, we can pull roles back to the light tier or switch a daily routine to weekly. Every one of those levers stays in your hands, in accounts you own.

One more thing we tell clients who ask whether paying for more credits lets us "build bigger." Usually no. The ceiling on what gets built is time and clarity, not tokens. More spend buys more volume of the same thing; a better role comes from a better brief. For how the fee itself works, see what does it cost, and for where all of this runs, see where does it run.