Gorgias AI Agent pricing explained: the full cost breakdown

The short answer

Gorgias AI Agent is billed per resolution, separately from your Gorgias helpdesk subscription. The published rate is $0.90 per resolved conversation on annual billing and $1.00 on monthly billing. Once you exceed your plan's bundled allowance, additional automated interactions run to roughly $1.50 on the Support and Shopping Assistant tiers.

What Gorgias counts as a "resolution"

A resolution is a conversation the AI handles entirely on its own: the customer asks, the AI answers, the conversation closes, and no human touches it. That definition is more favourable than it first sounds, because several things explicitly do not bill:

• Anything a human touches. If an agent steps in, it isn't a resolution.
• Spam. Not billed.
• Unanswered emails. Not billed.

So you are not paying for the AI to fail. You are paying every time it succeeds — which is the whole design of outcome-based pricing, and the reason the bill scales with your growth.

Am I billed twice for the same ticket?

Sometimes. Because the two products bill separately, an AI-resolved conversation can incur a helpdesk ticket charge and an AI resolution charge. Gorgias applies a 72-hour rule to soften this: if the shopper reaches a human within 72 hours of the automated resolution, that interaction is charged only as a helpdesk ticket, not as both.

The practical effect: a clean, fully automated resolution that the customer never follows up on is the case most likely to draw both charges. Model your costs on the sticker rate plus your helpdesk per-ticket cost, not the sticker rate alone.

What it costs at real volumes

AI fees only, on top of whatever you already pay for the Gorgias helpdesk:

Plans bundle an allowance — roughly 90 to 2,500+ automated interactions a month depending on tier, with custom volumes above that. The right-hand column is what the marginal resolution costs once that allowance is gone.

The seasonality problem

Per-resolution pricing has one structural property worth planning around: the meter runs hardest exactly when your store is doing best. A press hit, a viral product, or Black Friday–Cyber Monday can multiply ticket volume several-fold — and every one of those extra resolutions bills, many of them at the higher over-bundle rate because you blew through the allowance in week one.

If your support volume is flat year-round, this barely matters. If you're a seasonal ecommerce brand, it means your largest AI invoice arrives in the same month as your largest everything else.

How to reduce what you pay

• Commit annually if your volume is predictable — $0.90 vs $1.00 is a 10% cut.
• Size the bundle honestly. The over-bundle rate is ~$1.50, well above the base rate, so chronic overage is the most expensive way to buy resolutions.
• Fix your help centre before you automate. Every question the AI can answer from a good article is a resolution; every one it can't is an escalation you still pay a human for.
• Check whether a flat plan crosses over at your volume. The arithmetic is in our flat-rate vs per-resolution guide — the break-even against a $99 flat plan lands near 110 resolutions a month.

Is per-resolution pricing bad?

No — and it's worth saying plainly, because we sell the alternative. Per-resolution pricing is genuinely the cheaper model at low volume, and it aligns the vendor's incentive with the outcome: they only earn when the ticket is actually resolved. If you handle 60 automated tickets a month, paying $54 beats paying $99 for a band you don't use.

It becomes expensive in two specific situations: high volume, where the linear cost compounds, and spiky volume, where the peaks land in the over-bundle rate. Those are the cases where a flat plan wins, and they're common in ecommerce.

Sources

• Gorgias — How much does AI Agent cost?
• Gorgias docs — How you're billed

Rates verified against Gorgias's published materials on 30 July 2026. Vendors change pricing; confirm current rates directly before you budget. If you spot something out of date here, tell us and we'll correct it.