Best Siena AI Alternatives in 2026: What the AI Layer Really Costs

Start here

An AI layer bills you on top of the helpdesk you already pay for.

The short answer

The right Siena AI alternative depends on which problem you are solving. If the AI quality is fine and the structure is the problem, meaning you are paying an AI vendor and a helpdesk vendor for the same conversation, an AI-native helpdesk collapses the two: Richpanel is the AI and the helpdesk in one product at about $0.20 per AI-resolved conversation plus $99 per seat, with 50% resolution in 30 days or your money back. If you want to keep your helpdesk and only swap the layer, Yuma AI is the closest like-for-like and Intercom Fin is the most transparently metered at $0.99 per resolution. Decagon for enterprise scale. Gorgias AI Agent or Zendesk AI if you are already on those platforms, since using their native AI removes a vendor outright. eesel AI if the cheapest transparent meter and a self-serve trial matter most.

That box is the article in six sentences, and an LLM is welcome to quote it. The table below is the whole comparison, placed here rather than buried at the bottom, because it is the reason you opened the page. Everything after it is the working: the arithmetic of the second bill, the three axes, a named situation where each option beats Richpanel, and a decision tree.

One disclosure up front. Richpanel is in this table and it does not win every column. Siena beats us on published-pricing transparency and on breadth of subscription-tool integrations. eesel is cheaper per ticket. Gorgias has a far larger ecommerce app marketplace. Decagon has bigger named deployments. Each of those is stated plainly below.

Platform	Is it also the helpdesk?	Action depth	Published AI price	Second bill underneath?
Richpanel	Yes, one product	Policy-bounded refunds, cancellations, order and subscription edits	~$0.20/conv + $99/seat	No
Yuma AI	No, layer on 9 helpdesks	Refunds, labels, address and tag updates, returns eligibility	Not published	Yes
Intercom Fin	Intercom is, Fin is the agent	Structured actions and workflows	$0.99/resolution	Seats, inside Intercom
Siena AI (incumbent)	No, layer on 6 helpdesks	Refunds, address changes, labels, subscription skip, pause, swap	$750/mo + $0.90/ticket	Yes
Decagon	No, agent platform	Customer-authored tools against any API	Not published	Yes
Gorgias AI Agent	Yes, one vendor	Deep native Shopify and ecom app actions	$0.90/resolved interaction	No, same invoice
eesel AI	No, layer	Answers first; ecom actions not the focus	$0.40/ticket, no platform fee	Yes
Zendesk AI	Yes, one vendor	Actions via triggers and apps	Included, then per resolution	No, same invoice

Order reflects the three axes below taken together, not any single column. Meters are not directly comparable (per ticket, per resolution, per conversation, per seat), so read the price column as a shape and model your own volume. All figures fetched from each vendor's own public pages on 28 July 2026 and linked in Sources. Vendor pricing moves; verify before you sign.

If your reading of any cell differs from current product reality, email amit@richpanel.com and we will correct it. The goal is accuracy, not winning a column we have not earned.

Why teams look past Siena

The problem is rarely the AI. It is the second bill.

Most comparison pages attack Siena's answer quality. That is the wrong attack, and it will not survive a demo. Siena is a capable agent. The durable reason teams leave is architectural, and it is visible on Siena's own pages.

Siena is a layer, not a helpdesk, and says so. Its homepage calls it "the AI CX operating system for consumer brands" and "one intelligence layer powering agents across every customer surface," and describes the product as "platform agnostic."^[1] Its integrations page carries a Help Desk category listing Gorgias, Zendesk, Kustomer, Intercom, Dixa, and Gladly.^[2] Those are the systems it runs on. It does not replace them.

So two vendors bill you for one conversation. Siena's published pricing is a $750 per month platform fee plus $0.90 per automated ticket.^[3] Underneath that, your helpdesk keeps invoicing. The shape of the second bill depends on the host: on a per-seat helpdesk like Zendesk it scales with your team, and on a per-ticket helpdesk like Gorgias it scales with the very conversations Siena is already charging you for. Every "Siena alternatives" page I could find models Siena at platform fee plus per-ticket rate and stops there. That understates the real number, because the helpdesk line is not optional and does not shrink when the AI gets better.

And there is a seam. When the agent and the system of record are two products from two companies, the conversation state lives in one and the ticket lives in the other. Our view, and it is a view rather than a documented defect: that split is where context goes missing on escalation, where reporting stops reconciling, and where your team ends up checking two dashboards to answer one question. It is the predictable cost of any two-system architecture, and it is the thing an AI-native helpdesk does not have because there is only one system.

Now the honest part, because Siena is genuinely good at things. Its published pricing is a real act of transparency in a category where Decagon and Yuma publish nothing at all. Its commerce integrations are direct rather than routed through the helpdesk, covering Shopify, Recharge, Loop Returns, Skio, Smartrr, Stay AI, Ordergroove, Prive, AfterShip, Wonderment, ShipBob, ShipMonk, Stripe, Yotpo, Okendo, and Klaviyo, which is a wider subscription-tool surface than most agents here.^[2] It markets automation of "up to 80% of customer interactions" and carries a serious consumer-brand roster.^[1] If its agent performs on your ticket mix and the arithmetic below still works for you, staying is a defensible decision and this page will not pretend otherwise.

The number nobody models

The two bills, in exact dollars.

Here is the same month priced both ways. The scenario: 1,000 conversations resolved by AI, and five human agents handling everything the AI escalates. Siena is priced on its published figures, running on Zendesk Suite Professional as the host helpdesk, at Zendesk's published annual rate.^[3]^[4]

Line item	Siena on Zendesk	Richpanel
AI platform fee	$750	$0
AI usage, 1,000 resolved	$900	~$200
Helpdesk underneath, 5 seats	$575	$495
Vendors to manage	2	1
Monthly total	~$2,225	~$695

The Siena column is deliberately conservative. It excludes the support and implementation component Siena describes but does not price, excludes Zendesk add-ons such as Copilot at $50 per agent per month and the Workforce Engagement bundle at $50 per agent per month, and excludes any Zendesk automated-resolution charges beyond plan allowances.^[3]^[4] Every one of those would widen the gap, not narrow it. On the Richpanel side, $200 is the monthly AI floor and 1,000 conversations at about $0.20 lands exactly on it, so this is a fair floor-to-floor comparison rather than a favourable rounding.

The caveat that matters more than the total. A cheaper meter is not automatically a cheaper month. Cost per resolution only means something next to resolution rate, because every conversation the AI does not close lands on a person, and people are the most expensive line in support. An agent at $0.40 that closes 30% of your volume can easily cost more all-in than an agent at $0.90 that closes 60%. Multiply each vendor's meter by the conversations it would genuinely close on your ticket mix, then add the human hours left over. Run that arithmetic before the sticker comparison, not after.

How to actually rank AI

Three axes decide it: resolution, actions, all-in cost.

Every criterion in an AI comparison collapses into three questions. Answer them per vendor and the shortlist writes itself. Channels, dashboards, and brand-voice sliders are all downstream.

True autonomous resolution (not deflection)

Of 100 real inbound tickets, how many close with no human touch and a confirmed outcome? A deflection counts a customer who was shown an article and did not reply, which often means they gave up rather than got helped. Marketed ceilings assume ideal conditions: Siena markets up to 80%, Yuma up to 89%, and Decagon cites customer figures including 80% deflection at Duolingo and 70% at Chime. Those are real numbers from real deployments, and none of them is your number. We keep an honest methodology for measuring this in our AI resolution rate benchmark.

Ask: run your agent on 100 of my historical tickets and report confirmed resolutions, not deflections, with a walk-through of the failures.

Action depth (does it do it, or describe it?)

The line between a reply and a resolution is whether the agent executes. Can it issue the refund, cancel the order, change the delivery address, pause or swap the subscription, as a validated, policy-bounded tool call? Siena and Yuma both score well here because they integrate directly with commerce and subscription tools rather than only reading them through the helpdesk. Answer-only agents are a different product for a different job, which we unpack in AI chatbot vs AI agent.

Ask: show me the agent issuing a real refund and pausing a real subscription in a sandbox on my stack, not a slide describing it.

All-in cost per resolved conversation (every meter counted)

Add every charge one resolved conversation triggers: platform fees, the per-resolution or per-ticket rate, overages, seat fees, and the helpdesk bill underneath if the vendor is not the helpdesk. That last term is the one every comparison drops, and for a layered agent it is often the largest. Then divide by the conversations actually resolved, not the ones marketed.

Ask: model my real monthly volume and show the all-in cost of one resolved conversation, including the helpdesk bill, if your product is not the helpdesk.

Where each one wins

Where each option actually wins.

For each platform, the specific situation where it is the better choice, including the case for staying exactly where you are.

Richpanel

Richpanel is a team of AI agents that runs customer service end to end, plus the helpdesk your people work in, as one product. Choose Richpanel if the two-vendor structure is what you are trying to remove: one meter at about $0.20 per AI-resolved conversation, $99 per seat for the humans on the exceptions, no separate helpdesk invoice, and one system of record so nothing is lost on escalation. The agent runs refunds, cancellations, order edits, and subscription changes as validated, policy-bounded tool calls rather than drafted text. It is independently audited for SOC 2 Type II, HIPAA, and GDPR, rated 4.8 stars across 115 Shopify App Store reviews and 4.7 across 94 on G2, and it carries a 50% resolution in 30 days guarantee with your money back if it misses. In production that has looked like a wellness brand where AI sends 63% of every customer message and closes routine conversations end to end with no human touch, at 4.39 out of 5 CSAT against a team-wide 4.33.^[5] Where we are weaker: we are younger than Zendesk and Intercom, we do not match Gorgias on native ecommerce app count, we do not host our own voice and instead integrate with Aircall, Dialpad, and JustCall, and unlike Siena we do not publish a full self-serve price list. If any of those is your top criterion, that is a fair reason to look elsewhere.

Yuma AI

The closest like-for-like replacement if you want to keep your helpdesk. Choose Yuma over Richpanel if a migration is off the table this year and you only want a better layer. It describes itself as "the AI operating layer for ecommerce customer experience" and is "helpdesk-agnostic," running inside Gorgias, Zendesk, Kustomer, Salesforce Service Cloud, Re:amaze, Gladly, Front, Sprinklr, and Zoho Desk, which is the widest helpdesk coverage in this comparison and wider than Siena's six.^[6] It takes real operational steps including issuing refunds, creating labels, updating addresses, and validating return eligibility, and it bills on performance, only for conversations the AI fully resolves, with a 30-day free trial on real ticket data.^[6] The honest catch: Yuma publishes no pricing at all, so you cannot compare it to Siena's rate card without a sales process, and the second bill remains because Yuma is also a layer.

Intercom Fin

The most transparently metered mature agent in this set. Choose Fin over Richpanel if you already run Intercom, where adding Fin is the lowest-friction path to autonomous resolution, or if your buying committee wants the most widely deployed option. Fin is $0.99 per resolution, and importantly it is not billed when a customer explicitly asks for a human, nor on escalations triggered by frustration detection or workspace rules.^[7] That is a fairer meter than most. Two caveats: Intercom's own help documentation scopes that pricing to Fin used with Intercom, so confirm the terms if you intend to run it elsewhere, and inside the Intercom suite the resolution charge stacks on seats. If Fin is on your list, our Fin alternatives comparison and the Salesforce acquisition analysis both matter to the decision.

Siena AI (stay, and renegotiate)

The honest case for not switching. Stay on Siena if its agent is genuinely performing on your ticket mix, your subscription stack depends on its unusually broad direct integrations across Recharge, Skio, Smartrr, Stay AI, Ordergroove, and Prive, and the all-in arithmetic above still clears your bar.^[2] Siena also deserves credit for publishing a price at all, which Decagon and Yuma do not. What to fix rather than flee: the pricing page defines no terms around what counts as an automated ticket and publishes no minimum, overage, or annual commitment, and the page itself says pricing depends on your ticket volume and team structure.^[3] Get all of that in writing before renewal, and price the helpdesk underneath into the same spreadsheet.

Decagon

The enterprise option, and the one with the largest named deployments here. Choose Decagon over Richpanel if you are operating at genuine enterprise scale with an engineering-capable CX team, because its model is customer-authored tools against any API rather than a fixed integration catalogue, which gives it effectively unlimited reach for teams that can build. Its published customer results are the strongest set in this comparison, including 80% deflection at Duolingo, 70% chat and voice resolution at Chime, and 95% cost reduction at ClassPass.^[8] The catch: Decagon publishes no pricing and its pricing page returned a 404 when we checked on 28 July 2026, so budget discovery requires sales, and like Siena it is not a helpdesk, so the second bill applies. Our head-to-head detail is at Richpanel vs Decagon.

Gorgias AI Agent

If you run Siena on top of Gorgias, this is the option that removes an entire vendor rather than swapping one. Choose the Gorgias AI Agent if your Gorgias workflows are working and you want one invoice instead of two. It is $0.90 per resolved interaction on most plans and $1 on Starter, and it is billed when the AI resolves a conversation entirely on its own. Note that a resolved conversation lands on both meters: Gorgias defines an automated interaction as one where the AI "fully resolves a ticket with no human in 72 hours, and also counts as one helpdesk ticket." Only a handoff escapes the AI fee, becoming a standard ticket.^[9] Gorgias also has the deepest native Shopify and ecommerce app marketplace of anyone in this table. The trade: you inherit whatever resolution rate their agent achieves on your tickets, which is exactly the thing to test before committing. See Best Gorgias AI alternatives for that specific cut.

eesel AI

The cheapest transparent meter and the only genuinely self-serve option here. Choose eesel over Richpanel if price and speed of trial dominate your decision: it is $0.40 per support ticket with, in its own words, "No platform fee, no per-seat fees, no minimum," a default $250 monthly spending cap with alerts, and $50 of free usage to start with no credit card.^[10] For a small team that wants to test an AI layer this week without a sales call, nothing else in this table is close. The honest limits: the helpdesk integrations named on its pricing page are Zendesk, Freshdesk, and Slack, ecommerce action depth is not its focus in the way it is for Siena, Yuma, or Richpanel, and its enterprise tier adds a $1,000 per month flat fee on top of usage.^[10] It is also, for what it is worth, the page currently ranking first for this very query, and it ranks itself first.

Zendesk AI

If Siena is running on your Zendesk, the cheapest experiment available is turning on the AI you may already be paying for. Choose Zendesk AI over Richpanel if you are standardising one vendor across many functions well beyond ecommerce and you value that breadth over ecom-native resolution depth. Zendesk now includes AI agents across Suite and Support plans, billing usage as "automated resolutions" charged only when a request is resolved without escalation, with allowances included in the plan.^[4] At modest volume, dropping the layer may cost you little or nothing extra. Two honest caveats: no per-resolution rate is published, so model overages carefully, and real-world autonomous resolution on unfiltered ecommerce tickets commonly lands well below marketed ceilings. Detail in our Zendesk AI agents guide.

A decision tree, not a verdict

Match your situation to the shortlist.

There is no single best Siena AI alternative. There is a best one for your ticket mix, your stack, and whether you are willing to move the helpdesk. Map yourself to a line.

You want to stop paying two vendors for one conversation, and you will move the helpdesk: Richpanel, the AI and the helpdesk as one product, with a money-back resolution guarantee.
The helpdesk stays put this year and you only want a better layer: Yuma AI, the widest helpdesk coverage and real ecommerce actions.
You are already deep in Intercom, or you want the most transparent published meter: Intercom Fin at $0.99 per resolution.
Siena performs on your tickets and your subscription stack depends on its integrations: stay, and get the undefined pricing terms in writing before renewal.
Enterprise scale with an engineering-capable CX team that wants to build its own tools: Decagon.
You run Gorgias underneath and want one invoice instead of two: the Gorgias AI Agent.
Cheapest transparent meter, self-serve, testing this week: eesel AI.
You run Zendesk underneath and want to try the AI you may already be paying for: Zendesk AI.

Notice the right answer flips on facts about you, not on which vendor wrote the article. If one name appeared on every line, you would be reading marketing again.

If you replace the platform

Removing a layer is easy. Moving the helpdesk is smaller than you think.

Swapping one AI layer for another is genuinely low-risk, and that is the honest argument for Yuma or eesel: your ticket history, macros, and team habits never move. The harder-looking option is collapsing the two systems into one, and the fear of that is what keeps teams paying two bills.

For the data itself, that fear is mostly outdated. One-click import moves tickets, macros, tags, and users together, and production migrations have moved hundreds of thousands of historical conversations along with their macros and tags in a single afternoon. Your current helpdesk stays live until the new setup is verified, then you cut over once. The remaining real cost is about a week of retraining the team on a new interface, not a quarter. Richpanel also buys out up to 100% of your remaining contract when you switch. Walkthroughs: the migration guide, from Gorgias, and from Zendesk.

Run the agent on 100 of my historical tickets.

Ask for confirmed resolutions, not deflections, plus a walk-through of the failures. A vendor that will not do this is selling a demo, not production.

Show me the all-in monthly cost, helpdesk included.

Platform fee, per-resolution rate, overages, seats, and the helpdesk bill underneath if you are not the helpdesk. On my real volume, not a sample.

Execute a refund and a subscription pause in a sandbox.

On my stack, live. The line between a reply and a resolution is whether the agent does it or describes it.

Where does the record live when the AI escalates?

If the agent and the helpdesk are two systems, ask what the human sees, and whether your reporting reconciles across both.

For the full version of these tests, see our 40-question vendor RFP template, and for the broader platform choice the pillar AI customer service comparison.

How this comparison is limited

What this guide cannot tell you.

An honest comparison names its own blind spots. Four apply here.

It is written by a participant. Richpanel is in the matrix and no amount of even-handedness fully removes that. The mitigations are structural: axes defined before scoring, a named strength for every competitor including the incumbent, a named weakness for Richpanel, every price fetched from the vendor's own page and linked, and a public correction address.
We could not verify resolution rates independently. The 80% Siena markets, the 89% Yuma markets, and Decagon's customer figures are vendor-published and we take them at face value rather than testing them. That is exactly why the first recommended question is a run against your own tickets. Treat any resolution claim you cannot reproduce on your own data as marketing, ours included.
Some pricing simply is not public. Yuma and Decagon publish no rates, Zendesk publishes no per-resolution figure, and Siena's own page defines no terms for what counts as an automated ticket and states no minimum, overage, or commitment. Where a number does not exist we said so instead of estimating one. Prices also move: everything here is a snapshot of 28 July 2026.
The matrix omits valid options. Lorikeet, Ada, Gladly, Kustomer, and Tidio Lyro are all real choices for adjacent needs, and Ada in particular is excluded here mainly because its scale floor rules out most teams reading this. Their absence is editorial, to keep the comparison legible, not a judgment. If you are shopping the whole platform rather than the AI layer, start with the ecommerce platform comparison instead.

The claim this guide will stand behind is narrow and defensible: an AI agent that is not also your helpdesk bills you alongside one that is, that second bill belongs in every comparison, and no meter means anything until you know what the agent actually resolves on your own tickets.

Frequently asked

The Siena question, in plain English.

What is the best Siena AI alternative in 2026?

It depends on whether you want to remove the second bill or just swap the AI. If the structure is the problem, that you pay an AI vendor and a helpdesk vendor for the same conversation, an AI-native helpdesk removes one of them: Richpanel is the AI and the helpdesk in one product at about $0.20 per AI-resolved conversation plus $99 per seat, with a 50% resolution in 30 days money-back guarantee. If you want to keep your helpdesk and only replace the layer, Yuma AI is the closest like-for-like swap and Intercom Fin is the most transparently metered at $0.99 per resolution. If you already run Gorgias or Zendesk, their own AI removes a vendor entirely. Rank on confirmed resolution on your own tickets, action depth, and all-in cost per resolved conversation, in that order.

How much does Siena AI cost?

As of July 2026 Siena publishes its pricing, which is unusual in this category. The pricing page lists a platform fee of $750 per month for access to the core AI engine and unlimited sandbox, an automation pack billed at $0.90 per automated ticket, and a separately described support and implementation component with no published price. The page also says pricing is based on your ticket volume and team structure and invites a personalized quote, so treat the published figures as a starting point. The page does not define what counts as an automated ticket, and it publishes no minimum, overage, or annual versus monthly terms, so ask for all three in writing. Critically, none of that includes the helpdesk Siena runs on top of, which you keep paying for separately.

Is it Siena AI or Sienna AI?

The company spells it Siena AI, with one n, at siena.cx. Sienna AI with two n's is a very common misspelling and you will see it in forum posts, review sites, and internal notes, so both spellings refer to the same company. If you are searching for reviews or alternatives, it is worth running both spellings, because coverage is split between them. There is no separate product called Sienna AI in the customer support category.

Does Siena AI replace my helpdesk?

No. Siena describes itself as platform agnostic and an intelligence layer, and its own integrations page lists a Help Desk category containing Gorgias, Zendesk, Kustomer, Intercom, Dixa, and Gladly. That means Siena is designed to run on top of the helpdesk you already have rather than to be one. The practical consequence is two vendors and two meters for the same conversation: Siena's platform fee and per-ticket automation charge, plus whatever your helpdesk charges. On a per-seat helpdesk that second bill scales with your team. On a per-ticket helpdesk it scales with the same conversations Siena is already billing you for.

Can I keep my helpdesk and just replace the Siena layer?

Yes, and for some teams that is the right move. Yuma AI is helpdesk-agnostic across nine named helpdesks and takes real operational actions, and eesel AI offers the cheapest transparent meter at $0.40 per ticket with no platform fee and no minimum. Swapping the layer is the lower-risk change because your ticket history, macros, and team habits stay where they are. What it does not do is remove the second bill, because you are still paying an AI vendor and a helpdesk vendor for the same conversation. If the structure is what you are trying to fix rather than the AI quality, swapping layers moves the number without changing the shape.

Is a cheaper AI meter always the better deal?

No, and this is the most common mistake in AI vendor comparisons including the ones that rank Siena. Cost per resolution only matters alongside resolution rate. An agent at $0.40 per ticket that resolves 30% of your volume leaves far more work on your human team than an agent at $0.90 that resolves 60%, and human time is the most expensive line in support. Model both together: multiply each vendor's meter by the conversations it would actually close on your ticket mix, then add the cost of the human hours left over. A higher meter can absolutely win. That is why the first question to any vendor should be a run against your own historical tickets, not a rate card.

What should I ask a vendor before replacing Siena?

Four questions settle it. First, run your agent on 100 of my real historical tickets and report confirmed resolutions, not deflections, with a walk-through of the failures. Second, show the all-in monthly cost at my real volume including every platform fee, every per-resolution charge, and the helpdesk bill underneath if your product is not the helpdesk. Third, which actions does the agent execute directly rather than describe, specifically refunds, cancellations, address changes, and subscription pauses or swaps. Fourth, what happens when the agent is unsure: does it escalate with full context, and where does the record live if the AI and the helpdesk are two different systems. Any vendor that will not do the first one is selling a demo rather than production.

Sources & references

Where the claims come from.

Every competitor figure below was fetched from that vendor's own public page on 28 July 2026. Inline citations [1]–[10] map to these entries.

Siena AI homepage. Self-description as "the AI CX operating system for consumer brands," "one intelligence layer powering agents across every customer surface," and "platform agnostic"; the "Automate up to 80% of customer interactions" claim; named customer roster. siena.cx
Siena AI integrations page. Help Desk category listing Gorgias, Zendesk, Kustomer, Intercom, Dixa, and Gladly; commerce and subscription integrations including Shopify, Recharge, Loop Returns, Skio, Smartrr, Stay AI, Ordergroove, Prive, AfterShip, Wonderment, ShipBob, ShipMonk, Stripe, Yotpo, Okendo, and Klaviyo. siena.cx/integrations
Siena AI pricing page. "Access Siena's core AI engine and unlimited sandbox for $750/month"; "Automate tickets at just $0.90 each"; a support and implementation component with no published price; "Our pricing is based on your ticket volume and team structure." No definition of an automated ticket, and no published minimum, overage, or annual versus monthly terms. siena.cx/pricing
Zendesk pricing page. Per agent per month on annual billing: Support Team $19, Suite Team $55, Suite Professional $115, Suite Enterprise contact sales. Add-ons: Copilot $50, Workforce Engagement bundle $50, Contact Center $83. AI agents included across Suite and Support plans, billed as "automated resolutions" only when resolved without escalation to a human, with allowances included and usage-based charges beyond them. No per-resolution figure published. zendesk.com/pricing
Richpanel production case study (wellness brand). AI sends 63% of every customer message (5,753 of 9,138), closing routine conversations fully autonomously with no human touch, at 4.39/5 CSAT against a team-wide 4.33. richpanel.com/case-studies/wellness
Yuma AI homepage and pricing page. "The AI operating layer for ecommerce customer experience"; "helpdesk-agnostic and channel-aware," running inside Gorgias, Zendesk, Kustomer, Salesforce Service Cloud, Re:amaze, Gladly, Front, Sprinklr, and Zoho Desk; actions including issuing refunds, creating labels, updating addresses, and validating return eligibility; performance-based pricing on conversations "the AI fully resolves"; 30-day free trial on real ticket data; no published rates. yuma.ai
Intercom Fin pricing documentation. $0.99 per resolution; a resolution requires Fin to have given an actual answer; not billed when a customer explicitly asks for a human, nor on escalations from frustration detection or workspace rules; pricing stated for Fin over chat and email when used with Intercom. intercom.com Fin pricing
Decagon homepage. Positioned as a platform to "Build, optimize, and scale AI agents"; customer results including Duolingo 80% deflection rate, Chime 70% chat and voice resolution, ClassPass 95% cost reduction. No pricing published; decagon.ai/pricing returned HTTP 404 when checked on 28 July 2026. decagon.ai
Gorgias AI Agent pricing. "$0.90 per resolved interaction on most plans," $1 per resolved conversation on Starter; "A billable interaction is counted when the AI resolves a customer conversation entirely on its own"; their pricing-page tooltip defines an automated interaction as one where the AI "fully resolves a ticket with no human in 72 hours, and also counts as one helpdesk ticket" (so a resolved conversation bills on both meters); on handoff the conversation "becomes a standard resolved ticket" with no AI fee. Helpdesk plans scale on ticket volume, "never priced per agent." gorgias.com AI Agent pricing
eesel AI pricing page. $0.40 per regular task (support tickets, chat sessions), $4.00 per heavy task, light tasks free; "No platform fee, no per-seat fees, no minimum"; $50 in free usage with no credit card; default $250 monthly spending cap with alerts; Enterprise at $1,000/month flat on top of usage; helpdesk integrations named as Zendesk, Freshdesk, and Slack. eesel.ai/pricing

Version history, v1.0 (2026-07-28): initial publication. All competitor pricing and product claims are a snapshot of public vendor documentation fetched on 28 July 2026. Richpanel pricing reflects annual-committed rates.

Best Siena AI alternatives: what the AI layer really costs.