If you are shopping for AI support, the number that matters is not the sticker price on the plan. It is the cost per resolution: what you actually pay each time the AI handles a conversation, multiplied by how many conversations you have. That second part is the trap. Per-resolution pricing quietly ties your bill to your traffic, so the better your product does, the more you pay. This article is a calculator you can run yourself in two minutes.
We will use round, clearly labeled example numbers (these are illustrations, not anyone's quote) and walk the full arithmetic twice: once for a per-resolution fee, and once for bring your own key (BYOK), where you pay the model provider directly at cost with zero markup. By the end you will know the exact formula to drop your own real numbers into, and why the two models diverge as you grow. If you want the why behind the pricing first, read why AI support gets expensive.
The cost-per-resolution formula
Every AI support bill, no matter how it is dressed up, reduces to one of two formulas. Write both down, because the comparison is the whole point.
- →Per-resolution model: monthly cost = (conversations the AI handled) times (fee per resolution). The fee is fixed-ish; the count is whatever your traffic gave you.
- →BYOK model: monthly cost = (conversations the AI handled) times (tokens per answer) times (provider's per-token price). You pay the model provider at cost, and that is the entire bill for AI usage.
Notice that both formulas multiply by your conversation count, so both grow with traffic. The difference is the per-unit number you multiply by. In the per-resolution model it is a marked-up fee. In BYOK it is the raw cost of the tokens, which is usually a few cents or less. That single substitution is what changes the slope of the line.
Worked example: 2,000 conversations a month
Let us pick illustrative numbers. Say your widget handles 2,000 AI conversations a month. For the per-resolution side, assume an example fee of $0.99 per resolution (some tools charge around a dollar, some more, some sell buckets of conversations that work out to a similar per-unit rate). For the BYOK side, assume each grounded answer is roughly 3,000 tokens of input plus output (the visitor's question, a few relevant doc snippets, and a short reply) and an example blended provider rate that lands the cost around $0.02 per answer.
Now run both formulas at 2,000 conversations.
- →Per-resolution: 2,000 times $0.99 = about $1,980 per month, just for AI answers.
- →BYOK at cost: 2,000 times $0.02 = about $40 per month in tokens, billed straight to your own provider account.
Same 2,000 conversations. Same AI doing the same work. One bill is roughly $1,980, the other is roughly $40. The gap is not the AI getting better in one case. It is the markup sitting between you and the model in the per-resolution case, charged again on every single conversation. That is the entire story of why AI customer support is so expensive.
Now watch what happens when you grow
The 2,000 number was a snapshot. The real problem shows up over time, because per-resolution billing scales the marked-up unit, not the cheap one. Here is the same example at three traffic levels so you can see the slope.
- →500 conversations: per-resolution is about $495, BYOK is about $10.
- →2,000 conversations: per-resolution is about $1,980, BYOK is about $40.
- →10,000 conversations: per-resolution is about $9,900, BYOK is about $200.
Look at the per-resolution column climb. Every time your traffic multiplies, that bill multiplies with it, with no ceiling. The BYOK column climbs too, because tokens are not free, but it climbs at the cost slope instead of the markup slope, so even at 10,000 conversations it is a couple hundred dollars rather than five figures. The thing you want most, more people using your product, is exactly the thing per-resolution pricing punishes hardest.
Build your own calculator in a spreadsheet
You do not need a fancy tool to run this on your real numbers. Open a spreadsheet, drop in four cells, and you have a cost-per-resolution calculator that beats any vendor's marketing page.
- 01Cell A: your monthly AI conversations (estimate from your current chat or ticket volume).
- 02Cell B: the per-resolution or per-conversation fee the tool charges. If it sells buckets, divide the bucket price by the bucket size to get a per-unit number.
- 03Cell C: your estimated token cost per answer. A safe starting guess is a few cents; check your provider's per-token price and multiply by roughly 3,000 tokens.
- 04Result: per-resolution monthly = A times B. BYOK monthly = A times C. Compare the two, then re-run them at 2x and 5x your current traffic to see your future bill.
If the BYOK column is dramatically smaller and stays smaller as you scale the traffic up, you have just proven the case for paying the provider at cost. That is the whole pitch of bring your own key: you delete column B entirely and live on column C.
How muro prices this, honestly
We built muro so the per-resolution column never exists. There is no per-resolution fee and no per-conversation meter that grows with your traffic. You run the AI one of two ways, and we will be straight about both.
- →BYOK. Paste your own Anthropic key and the AI Answer Agent runs unlimited, at cost, with zero markup. You pay the provider directly for tokens and nothing extra to us for AI usage. This is column C with column B removed, and it is what we point heavy-traffic users to.
- →Managed credits. Do not want to manage a key? Every plan includes a monthly pool of AI credits where one credit is one action, and you can buy a top-up pack in a big month. It is metered, but it is a predictable pool you choose, not an open-ended per-resolution charge that climbs with your traffic.
The rest of muro is flat too. Solo is $19/mo for two projects and Fleet is $59/mo for unlimited projects, both with a 14-day trial and no card required. None of those prices move because your support got busy. See the full flat pricing, or try it on a live demo first.
Install the widget and start measuring
The fastest way to get real numbers for your calculator is to put the widget on your site and watch your actual conversation volume for a week. It is one tiny script (about 6 KB) and one line to initialize. Paste this before the closing body tag and replace YOUR_WIDGET_ID with the id from your muro dashboard.
html<script>
(function(w,d,s,o){
w.MuroChat=o;w[o]=w[o]||function(){(w[o].q=w[o].q||[]).push(arguments)};
var j=d.createElement(s);j.async=1;j.src='https://muro.chat/widget.js';
d.getElementsByTagName('head')[0].appendChild(j);
})(window,document,'script','muro');
muro('init',{widgetId:'YOUR_WIDGET_ID'});
</script>Once it is live, your dashboard shows how many conversations the AI actually handled. Plug that real number into cell A of your spreadsheet and the calculator stops being hypothetical. If you turn on BYOK, your provider account shows the real token spend, so column C is real too. No estimates, no guessing, no marked-up meter in the middle.
Cost per resolution is just a multiplication: conversations times your per-unit cost. Per-resolution billing makes that per-unit number a markup and charges it again on every conversation, so your bill scales with your traffic and your best months hurt the most. Pay the provider at cost instead and the slope flattens. With muro you can bring your own Anthropic key and run AI unlimited at zero markup, or use a predictable credit pool with no per-resolution fee. Start free, no card required, and run the numbers on your own traffic.