Gorgias AI Agent, Siena AI, Rep AI and Yuma AI all promise to close ecommerce support tickets for you. Their headline rates look similar — roughly a dollar a ticket. What differs is what they count before they charge you, and what you still pay for the helpdesk underneath. That is what this guide compares.
Per automated interaction, on top of a ticket-priced helpdesk plan from $10/mo. Bundled AI allowances apply; overage is $1.50 each.
Flat platform fee plus a per-automated-ticket rate. Final pricing is quoted after you share your volume — implementation is not publicly priced.
Website plans scale with monthly sessions, not resolutions. Example: Concierge from $130/mo at the 10,000-session level, on published figures.
Annual bundle for 12,000 full resolutions (about $1.20 each), paid upfront on a 12-month commitment. Larger bundles lower the per-resolution rate.
| Gorgias AI Agent | Siena AI | Rep AI | Yuma AI | |
|---|---|---|---|---|
| Best for | Shopify stores that want the helpdesk and the AI in one product | DTC brands that want an AI CX layer without changing helpdesk | Shopify stores where pre-purchase selling matters as much as support | Higher-volume DTC teams that want the strictest definition of a billable resolution |
| What you are billed for | Each automated interaction: a ticket the AI resolves alone | Each “automated ticket” — see the definition warning in the next row | Monthly website sessions (traffic) for the on-site plans; resolutions only on the email / social add-on | Each “full resolution”: a ticket handled end to end with no human involved |
| When you are NOT billed the AI fee | When a human replies within 72 hours, the automation fee does not apply (the helpdesk ticket fee still does) | Not clearly defined on Siena's public pricing page. Competitor comparison pages describe Siena as charging per conversation whether or not it resolves — get the definition in writing before you sign | There is no per-resolution fee to avoid on website chat; the traffic tier is the bill. The email / social / WhatsApp add-on is listed at $0.75 per AI-resolved conversation | Escalations, partial handling and repeat actions on the same ticket are not charged, per Yuma's pricing page |
| Helpdesk underneath | It is the helpdesk — but you pay for both meters: an AI-resolved ticket also counts as a billable helpdesk ticket | You keep (and keep paying for) your own helpdesk — Siena is a layer on top | Can hand off to Gorgias or Zendesk, or use Rep's own inbox product, priced separately | You keep (and keep paying for) your own helpdesk — Yuma is a layer on top |
| Published price shape | Helpdesk $10–$900/mo by ticket volume, plus $0.90 (annual) / $1.00 (monthly) per automated interaction; plans bundle 30–530 interactions, then $1.50 each | $750/mo platform fee + $0.90 per automated ticket; support and implementation not publicly priced; final price quoted per store | Session tiers: e.g. Support from $49/mo and Concierge from $130/mo at 10,000 sessions; $726/mo for Concierge at 80,000 sessions in published snapshots; extra sessions $12 per 1,000 | Annual bundles paid upfront: $14,400/yr for 12,000 resolutions (~$1.20 each), $40,000/yr for ~48,000 (~$0.83), $90,000/yr for ~120,000 (~$0.75) |
| Commitment shape | Monthly or annual helpdesk plan; annual billing lowers the AI rate | Sales-led quote; no public free plan | Monthly session tiers with a 14-day trial on published terms | Every plan carries a 12-month commitment; annual plans pause at the bundle limit instead of billing overage |
| Trade-off to know | Two meters stack on the same ticket, and resolutions past your bundle cost $1.50, not $0.90 | The $750/mo floor applies before the AI resolves anything, and the billable unit is the least clearly defined of the four | You pay for traffic, including visitors who never chat — cheap in a busy month of resolutions, fixed cost in a quiet one | Highest published entry commitment of the four: $14,400/yr upfront, before your helpdesk bill |
Prices are USD, before tax, as published by the vendors in November 2026. Gorgias's own materials state the AI rate two ways — $0.90 on annual / $1.00 on monthly billing, and a per-plan table running from $1.00 (Starter) to $0.85 (Advanced) — so confirm the exact rate for your plan and billing cycle. Rep AI's pricing page is a traffic slider: the price you see depends on the session volume selected, and figures in third-party listings differ for that reason. Plans change — always confirm the current price, the billable-unit definition and what is included on the vendor's site before you buy.
Headline rates cannot be compared directly, so here is one store priced four ways. The assumptions are ours, stated plainly so you can redo the maths with your own numbers: 2,000 support tickets a month, of which the AI fully resolves 50% (1,000 a month, 12,000 a year). The 50% figure is an assumption for the calculation, not a vendor claim and not a prediction for your store. Where a product needs a separate helpdesk, we price the layer over a Gorgias Pro helpdesk on annual billing ($300/mo, 2,000 tickets included) as a representative example — substitute your own helpdesk cost if you run something else.
| Product | How the example bill is built | Total for the example store |
|---|---|---|
| Gorgias | Helpdesk Pro $300/mo (annual). AI: 190 bundled interactions at $0.90 = $171, plus 810 resolutions past the bundle at $1.50 = $1,215. AI subtotal $1,386/mo | $1,686/mo · $20,232/yr |
| Siena AI | Platform fee $750/mo + 1,000 automated tickets × $0.90 = $900, plus the separate helpdesk underneath at $300/mo | $1,950/mo · $23,400/yr — if “automated ticket” means a resolved ticket |
| Siena AI, the other reading | Same store, but if billing follows the per-conversation reading used in competitor comparisons — the AI touches all 2,000 conversations and resolves 1,000: $750 + 2,000 × $0.90 + $300 helpdesk | $2,850/mo · $34,200/yr — a $10,800/yr gap created by one definition |
| Yuma AI | Starter bundle $14,400/yr covers exactly the example's 12,000 resolutions, plus the separate helpdesk at $300 × 12 = $3,600/yr | $18,000/yr ($1,500/mo equivalent) — but the AI bundle is paid upfront |
| Rep AI | Cannot be priced from ticket counts at all: the on-site Concierge price follows store traffic. At the published 80,000-sessions level it is $726/mo, whether the AI resolves 500 chats that month or 1,500 | $726/mo · $8,712/yr for on-site chat at that traffic level, before any email / social add-on or inbox costs |
Three lessons from the arithmetic. First, the cheapest-looking rate is not the cheapest bill: Gorgias's $0.90 headline becomes a higher effective rate once resolutions pass the bundled allowance at $1.50. Second, the definition is money: the two Siena readings differ by $10,800 a year on identical ticket volume, which is why the billable event — not the rate — is the first thing to pin down in any contract. Third, compare against your own human cost the honest way: divide your support team's total monthly cost by the tickets they handle. If, purely as an illustration, that works out at $5 of staff cost per ticket, then an AI resolution priced around $0.90–$1.20 plus its share of platform and helpdesk fees is the comparison to make — remembering that the AI tends to take the routine tickets first, and the tickets left for your team are the slower ones.
Gorgias is an ecommerce helpdesk first: email, chat, social, SMS and voice in one inbox, with Shopify order data — refunds, cancellations, order edits — inside the ticket. Its AI Agent (the automation product formerly presented under the Automate name) answers and acts on routine post-purchase questions such as order status, returns and product queries. The helpdesk is priced by billable tickets, not seats: Starter $10/mo for 50 tickets, Basic $60/mo for 300, Pro $360/mo for 2,000 and Advanced $900/mo for 5,000 on monthly billing, with annual billing bringing Basic to $50, Pro to $300 and Advanced to $750 a month. The AI Agent is a second meter on the same invoice: $0.90 per automated interaction on annual billing, $1.00 monthly, charged only when the AI resolves a conversation with no human involved within 72 hours. Each plan bundles some interactions (30 on Starter and Basic, 190 on Pro, 530 on Advanced); beyond the bundle each one costs $1.50. Gorgias itself describes the automation potential as up to around 60% of conversations — a vendor figure, not an independent measurement. The structural point buyers miss: an AI-resolved ticket still counts toward your helpdesk ticket allowance, so one conversation can legitimately appear on both meters.
Siena positions itself as an AI customer-experience operating system for consumer brands: an intelligence layer that runs on top of the helpdesk you already have — its published integrations include Gorgias, Zendesk, Intercom, Kustomer, Gladly and Dixa — and handles support conversations in your brand voice, with order actions such as cancellations, refunds and address changes where the underlying setup allows. Its public pricing page is short: a $750/month platform fee covering the core AI engine and an unlimited sandbox, an Automation Pack at $0.90 per automated ticket, and a support-and-implementation tier whose price is not published. There is no free plan and no self-serve plan table; the page asks for your ticket volume and team structure before quoting accurately. Two things deserve a buyer's full attention. The $750 floor is payable before a single ticket is automated, and it stacks on top of your existing helpdesk bill. And Siena's own page never defines precisely what makes a ticket “automated” for billing — while comparison pages published by competitors describe Siena as charging per conversation regardless of outcome. Those competitor descriptions are not neutral sources, but the gap itself is the finding: this is the one product of the four where you cannot reconstruct the bill from public information alone.
Rep AI is the exception in this comparison, and that is exactly why it belongs in it. Built for Shopify, its agent combines a sales side — behavioural triggers that open a chat when a shopper hesitates, product recommendations, adding items to the cart inside the conversation — with a support side covering order tracking, returns, exchanges and cancellations, handing off to Gorgias or Zendesk when a human is needed. Its website plans are priced by monthly store sessions, not by chats, replies or resolutions: published figures put Support from $49/mo and the combined Concierge from $130/mo at the 10,000-session level, rising to $726/mo for Concierge at 80,000 sessions, with extra sessions billed at $12 per 1,000. A busy month in which the AI resolves twice as many chats costs the same as a quiet one at the same traffic level — and a quiet month costs the same too. One part of Rep does use outcome pricing: AI for email, social and WhatsApp is offered as an add-on listed at $0.75 per AI-resolved conversation. A caution on the numbers themselves: Rep's pricing page is a slider set by your traffic, so starting prices quoted around the web refer to different session levels and differ from each other — always read the price at your own store's traffic.
Yuma is an AI agent layer for ecommerce brands that sits inside the helpdesk you already run — published helpdesk support includes Gorgias, Zendesk and Kustomer among others — and works tickets end to end: reading the ticket, pulling live order and customer data, checking your policies, replying in your brand voice and performing the closing action, such as a refund, a return label or a subscription change. Its pricing page is built around a single defined term, the Full Resolution: a ticket Yuma handled from start to finish with no human involved, counted once per ticket. Conversations, escalations and repeat actions on the same ticket cost nothing, and any human touch removes the charge. The catch is the payment shape: plans are annual bundles paid upfront on a 12-month commitment — Starter at $14,400/year for 12,000 resolutions (about $1.20 each), Growth at $40,000/year for roughly 48,000 (about $0.83), Scale at $90,000/year for roughly 120,000 (about $0.75), with the rate falling toward $0.55 at higher volumes. Annual plans do not bill overage — usage pauses when the bundle runs out — while monthly plans charge $1.50 per extra full resolution. Yuma publishes no implementation fee and offers the first 30 days free on its pricing page. Its published case material reports automation rates that vary widely by brand, including a peak-season figure of 89% for one named retailer; treat those as vendor-reported examples for specific stores, not as an expected rate for yours. Note also that older third-party pages carry outdated Yuma rates — the annual bundles above are the current published structure.
Before rates: what exactly triggers a charge? A full resolution with no human touch (Yuma, and Gorgias with its 72-hour rule), an “automated ticket” you should ask the vendor to define (Siena), or your traffic level (Rep). The worked example above shows one undefined term swinging a bill by $10,800 a year.
Every total in this guide includes a platform layer. With Gorgias the second meter is on the same invoice; with Siena and Yuma your existing helpdesk bill continues in full; with Rep, decide whether its inbox or a handoff to your current desk is part of the setup. Never budget the AI rate alone.
Vendor automation figures are reported on their own customers' ticket mixes. Count your own last 1,000 tickets: how many are order-status, returns and other routine intents the AI can plausibly close end to end? Price the bundles and overages at that rate — and at half of it — before committing to an annual bundle.
Per-resolution metering (Gorgias) flexes with volume but stacks two meters. A platform fee (Siena) is a fixed floor. Traffic pricing (Rep) is stable but never sleeps. An upfront annual bundle (Yuma) is predictable per resolution but punishes over-forecasting. The right shape depends on how seasonal your store is.
It means the vendor charges when the AI closes a conversation, rather than a flat monthly fee or a charge per message. But “closes” is defined by each vendor: Gorgias charges its automation fee when the AI resolves a ticket with no human involved within 72 hours; Yuma charges only for a ticket handled end to end with no human involved at any point; Siena's public page says “automated ticket” without defining the trigger precisely; and Rep AI does not use per-resolution pricing for website chat at all — it charges by store traffic, with per-resolution pricing only on its email and social add-on. The definition, not the headline rate, decides what you pay.
Not necessarily on rate: at $0.90–$1.00 per automated interaction, Gorgias sits in the same band as its rivals. The surprise is structural. An AI-resolved ticket also consumes a billable helpdesk ticket, and resolutions beyond your plan's bundled allowance cost $1.50 each, so the effective average rate rises with your automation volume. Stores shopping for an alternative are usually reacting to that stacking — or to wanting an AI layer that works on a helpdesk they already run, which is the Siena and Yuma model.
It depends on the product. Gorgias is the helpdesk, so adopting its AI Agent means adopting (or already using) Gorgias. Siena and Yuma are layers designed to run inside an existing helpdesk such as Gorgias, Zendesk or Kustomer — you keep your desk and pay for both. Rep AI works in front of Gorgias or Zendesk, or with its own inbox product. In every case, budget the combined stack, not the AI line alone.
Treat them as vendor-reported figures from specific customers, not as expectations. Gorgias describes automating up to around 60% of conversations; Yuma's case material spans a wide range across brands, with one named retailer reported at 89% in peak season. Your rate will depend on how much of your queue is routine, policy-clear work. The safer budgeting method is the one in the worked example: assume a rate, state it, and check the bill still makes sense at half that rate.
Run the floors, not the rates. Siena's $750/mo platform fee and Yuma's $14,400/yr entry bundle are sized for substantial volume — at 200 resolutions a month, the platform fee alone would add several dollars to every resolution. Gorgias's entry tiers and Rep's low-traffic session tiers publish much smaller starting points. At low volume, also price the boring alternative honestly: your own team's cost per ticket, using the division in the worked example.
Because this guide is compiled from publicly available vendor information, not hands-on testing. We have not run these agents against a live store's ticket queue, so we do not publish resolution-rate scores or first-hand impressions. Where a judgment appears — for example, that an undefined billing term is a risk — it follows from the published pricing documents themselves, quoted next to it so you can judge for yourself.