Decagon pricing in 2026: what a contract costs

Decagon does not publish a price. Its pricing page returns a 404, and we found no free trial or self-serve sign-up, so the sales demo is the only way in. Prices on this page are in US dollars.

Vendr’s data is the best we found: 12 contract documents from two buyers, with a median contract of $432,750 a year and a displayed range of $105,000–$923,183 a year.¹ That sample is thin enough to move at the next check. The two figures other guides repeat for Decagon, a platform fee and a rate per conversation, rest on less. We trace both below, then use them to price a month of conversations.

How Decagon bills

Decagon states two ways to buy, and says which one most customers choose. Its Director of Product wrote in December 2024 that per-conversation pricing is “a fixed rate for every incoming conversation, with flexible pricing for higher volumes”, and that per-resolution pricing is “a higher fixed rate for each fully resolved conversation, with no charge for escalations”.² The same post adds that “the vast majority of our customers choose per-conversation pricing.”

Under the default, every conversation the AI touches is billed, including the ones it hands to a person. Per resolution reverses that: escalations are free, and the rate per resolution is higher. Decagon gives no number for either.

Nothing else about the bill is stated on decagon.ai, so there is no rate, no minimum and no platform fee to quote. Its AWS Marketplace listing, written by Decagon as the seller, has one line, “Platform Access”, with “Pricing determined by private offer”.³ That cannot tell us whether a separate platform fee exists. Vendr’s listing describes the same shape, per conversation and varying with volume, on Net 30 or Net 60 terms, and its catalogue has a “Platform Fee” line with no amount shown.

What buyers pay

The highlighted row is the one most guides quote; read the source column first.

The Vendr rows need the sample in mind. Twelve documents from two buyers are not twelve independent prices. Vendr’s chart of contract sizes starts at $100,000 a year, which is the cheapest contract in it; the $105,000 at the bottom of the displayed range is a little above it, and is not the cheapest contract.

The only buyer we found talking about price is on Vendr, and the phrase “list price” in the note is the one sign that Decagon has a list price at all.

“We were able to get a 30% discount off list price for 1mil+ conversations on our new purchase by emphasizing budget constraints and focusing on the budget required in order for leadership to sign off.”

A buyer at a company of 201–1000 staff, on Vendr, 30 August 2024¹

The same buyer, or another company of the same size, also reports unlimited admin seats, a two-year deal with a year-two opt-out, quarterly instead of annual payment, and SLA credits.

The platform fee and the rate per conversation

The two numbers most Decagon guides repeat are a platform fee of about $50,000 a year and a rate of about $0.99 a conversation. Fin’s and Quiq’s guides both quote them,^{4, 5} and we could not trace the amounts to a contract, a buyer or Decagon.

The sourcing leads nowhere. Fin credits the fee to eesel, Quiq and Featurebase. Featurebase’s page has no dollar figure for Decagon and says Decagon “doesn’t publish list pricing”.⁶ On eesel’s current page, $50,000 is Vendr’s minimum redline threshold, and a separate line says contracts “typically start around $50,000/year based on third-party estimates”, with no link.⁷ Quiq credits “several sources and reviews” and links only to its own blog.

Vendr’s own page does carry a figure of $50,000, but it is a different kind of number: a machine-generated estimate tied to legal redlines on a contract, not a charge. We think the $50,000 figure began as that threshold, misread and then copied from blog to blog. We cannot prove it.

The rate has a problem of its own. Fin attributes it to “one third-party source” and Quiq to “public information and industry discussions”; neither names the origin. It is exactly Fin’s published price per resolution, and one of the two pages quoting it for Decagon is Fin’s. That proves nothing alone, but it is a reason to hold the number loosely. Fin’s page also gives $0.50 a resolution for Decagon’s per-resolution option, which sits below the per-conversation figure, the reverse of what Decagon states, so we dropped it.

What the quote leaves out

Above the minimum, every conversation the AI does not resolve is still billed. At 20,000 conversations a month with 60% resolved, the 8,000 it does not resolve cost about $7,920 at the quoted rate, and the people who take them still need a helpdesk. Decagon hands conversations to Salesforce, Zendesk and Intercom,⁸ so those licences continue.

Nothing Decagon publishes settles the payment terms. A Vendr buyer note says “Annual payment terms are standard”, while the listing’s own field says “Monthly”. Decagon’s refund policy on AWS Marketplace says “All fees are non-cancellable and non-refundable except as required by law.” We could not read Decagon’s master services agreement: decagon.ai/terms redirects to it, but the page returned no legal text, so its fee and renewal terms are unchecked.

Setup takes staff time. Decagon’s AWS listing says enterprises deploy in six to seven weeks, and reviewers describe tuning that goes on after launch.

Before a proposal arrives, ask Decagon whether a platform fee sits apart from usage, what the rate is at your volume and where it steps down, whether an escalated conversation is billed at the full rate, and what the renewal terms say. We found no implementation fee stated anywhere, so ask about that too.

What 3,000 conversations a month costs

At small volume, Decagon’s price is a minimum contract, so the calculator has a floor as well as a rate. Its inputs are the unverified fee and rate from the section above, with the fee counted on the high side only, and Vendr’s cheapest contracts as the floor. Decagon’s side does not move with the share the AI resolves, because every conversation is billed; that slider changes Ezro’s side only.

At the default, the minimum sets the bill. On the low side, 3,000 conversations at $0.99 come to $2,970 a month, against $8,333 for a $100,000 contract spread over twelve months. On the high side, the $50,000 fee spread monthly is $4,167, and $4,167 plus $2,970 is $7,137, which is under the $8,750 a month that a $105,000 contract comes to. Either way, at 3,000 conversations you pay the minimum, about $100,000 a year.

Without a platform fee, usage passes the minimum at about 8,400 conversations a month ($8,333 divided by $0.99). With the $50,000 fee, it passes it at about 4,600 ($8,750 minus $4,167, divided by $0.99). At 30,000 conversations the range is $29,700–$33,867: 30,000 at $0.99 is $29,700, and the fee adds $4,167. At the same rate, the median contract is $36,063 a month, which buys about 36,400 conversations ($36,063 divided by $0.99).

Who it’s right for

Decagon is the better buy for a company with high, steady volume that wants voice, chat, email and SMS from one vendor. Decagon says one agent covers all four, with simulations, live A/B tests it calls Experiments, and a monitoring layer called Watchtower that checks every conversation. Those are the vendor’s claims, and we have not tested them.

Reviewers are positive, and specific. G2 reviewers, syndicated on AWS Marketplace, rate Decagon 4.8 out of 5 across 29 reviews,⁹ and the praise is concrete. “I’m also impressed by how easy it is to scale support during busy periods,” writes Akash R. (3 September 2026). Adam R. (21 July 2026) says “our CX team can turn complex workflows into simple English and then update refunds and troubleshoot issues more precisely.” Decagon is growing fast: in January 2026 it announced $250 million at $4.5 billion, and said it had added more than 100 new global enterprise customers in the year.¹⁰

The complaints are about setup effort and handling harder questions, and none we found is about price. A reviewer in logistics and supply chain (28 July 2026) says Decagon “takes some time to set up and fine-tune everything properly”. Akash R. writes that “Sometimes the AI needs additional context or a human agent to step in”. Rahul B. (20 August 2026) goes further on accuracy: when problems show up in company policies, the resolutions Decagon provides “might not be proper and correct”. The same review calls the setup “quite complex”. The reviews are all four or five stars and many read alike, so we weight them lightly.

Where Ezro is different

Ezro is an AI operator. You write down how your team does a job, and Ezro follows that procedure in your systems, across email, Instagram, Messenger and WhatsApp. Decagon’s Agent Operating Procedures are written in plain language too, so the idea is close and the buying is where we differ. Our prices are published: plans start at $99 a month, on Starter and Growth a resolution past the allowance costs $0.40 to $0.50, and a plan counts resolutions, not conversations.

Questions

Ezro’s prices are on one page →