Decagon pricing
Decagon does not publish a price. Its pricing URL returned a 404 on 29 September 2026. Its own post describes two models, per conversation and per resolution, and says most customers choose per conversation. This page states what that post makes public, what stays a quote, and how that compares with vendors that print a rate.
A model, not a rate card.
Decagon's public writing explains how it likes to bill. It does not say what a conversation costs. On 29 September 2026, decagon.ai/pricing returned a 404. The pricing post that does exist names two options and says the vast majority of customers choose one of them. There is no self-serve checkout and no printed minimum.
Per conversation is a fixed rate for every incoming conversation, resolved or not, with flexible pricing at higher volumes. Per resolution is a higher fixed rate for each fully resolved conversation, with no charge when the conversation escalates. Larger resolution commitments lower that rate. Neither option is given a dollar figure in the post.
Decagon's argument for the conversation model is predictability. Outcome definitions get argued: a customer who leaves, a trivial answer, a partial fix. The post says most customers would rather not have that argument, and that a conversation rate is easier to budget. That is Decagon's claim about its own book of business, not a survey we ran.
Third-party pages quote dollar amounts for Decagon. Decagon has not published those amounts. This page does not repeat them as a list price. If you need a number before a call, Decagon's site will not give you one.
The two models Decagon describes
Both rows are from Decagon's own post. The dollar rate on either model is not in that post, and it is not on a pricing page.
| Per conversation | Per resolution | |
|---|---|---|
| What is billed | Every incoming conversation | A fully resolved conversation |
| Escalations | Billed at the full rate | Not charged |
| Relative rate | The lower of the two | Higher, Decagon says |
| Volume | Flexible pricing at higher volume | Larger commitments lower the rate |
| Who chooses it | Most customers, Decagon says | Offered, not the default |
| Dollar figure | Not published | Not published |
What you can price before a call
Fin and Sainapse print numbers a buyer can multiply. Decagon's public writing describes the shape of the bill, not the rate.
| Decagon | Fin | Sainapse | |
|---|---|---|---|
| Published dollar rate | None | $0.99 per outcome | From $500 a month |
| Unit | Conversation or resolution | A defined outcome | An AI-handled ticket |
| Escalations | Charged on the conversation model | No outcome, no charge | Not a per-outcome meter |
| Where to read it | A blog post, not a rate card | fin.ai/pricing | The Sainapse pricing page |
| Before a call | A quote | A trial and a published rate | The public plan fees |
| Minimum on the page | None published | 50 outcomes a month | 1,000 tickets on Starter |
What is public, what is not
When Decagon's pricing is the better fit
Decagon is the better buy when you want a consumer-scale concierge and you would rather pay a predictable rate per conversation than debate what counted as a resolution. That is Decagon's own case for the model most of its customers pick. If that argument matches how your finance team budgets, the missing rate card is a procurement step, not a reason to disqualify them.
Decagon is also the better buy when you want the vendor to own the agent build for chat and voice, and you accept that the price arrives in a quote. A webpage that names the model and withholds the rate is consistent with that sale. It is not a hidden fee. It is an unpublished one.
Fin is the better buy when you need to multiply a rate by volume before anyone joins a call. $0.99 per outcome, a 50-outcome minimum, and a 14-day trial are on Fin's pricing page. Read the outcome list before you multiply. A qualification is $9.99, not $0.99.
Sainapse fits when the desk stays and you want a platform fee with included AI-handled tickets, plus a published overage on Starter and Growth, instead of a quote for conversations. The units are different. A lower line on one page is not proof the other bill would have been higher.
What Sainapse runs in production
Ford, four-plus years live
300+ specialists, 250K+ tickets a year, embedded natively in ServiceNow. 95% copilot accept rate. Resolutions applied by engineers without edits. ~100% triage and routing accuracy. Hand-offs per resolved ticket 4 to 1.2.
An enterprise SaaS support desk
400+ specialists, 1.2M+ tickets a year, 12 months live inside the helpdesk. Customer-initiated touchpoints per resolved ticket 4 to 1.2. CSAT +35%, concurrent with the Sainapse period, as correlation, not a causal claim.
How these numbers are measured
Accept rate is the service desk's own accept and reject telemetry, not a survey. Hand-off counts exclude automated steps. Satisfaction movement is published as correlation because that is what the readout is.
Sources and retrieval dates
Common questions
Decagon does not say. Its pricing URL returned a 404 on 29 September 2026, and its pricing post names two models without a dollar rate. Third-party contract estimates circulate. None of them are confirmed on Decagon's site, so this page does not print them. The number you can defend in a budget meeting is the one in the quote.
Per conversation, according to Decagon. The post says the vast majority of customers choose a fixed rate on every incoming conversation, including ones the agent escalates. Per resolution costs more per success and does not charge the escalation. Which one a given contract uses is still a sales conversation.
Nobody can show that from public pages. Fin publishes $0.99 per outcome. Sainapse publishes plans from $500 a month and a $0.50 overage on Starter and Growth. Decagon publishes neither a rate nor a minimum. The units are also different, so a spreadsheet that subtracts them is not a comparison.
A plan fee, an included volume, and an overage. Starter is $500 a month with 1,000 AI-handled tickets. Growth is $2,000 a month with 5,000. Extra tickets on those two plans are $0.50, under a cap you set. Enterprise is a custom annual commitment, not a printed overage. That is still not a Decagon quote.
When the deployment is the product: a vendor-led concierge for consumer chat and voice, and finance will price it in a contract rather than on a webpage. Decagon is built for that sale. If you will not start without a public rate, Fin and Sainapse both print one, and they finish different work.
Where these claims come from
Each link backs a price, a definition, or a sibling comparison named above.
- Decagon alternatives
Who to shortlist when the missing price is only one of the objections.
- Sierra pricing
The other quote-only agent: outcome-based, and no dollar rate published.
- Intercom Fin pricing
The published $0.99 outcome rate, itemized, including what is not $0.99.
- Fin vs Decagon vs Sierra
Pricing model, stack, and fit for the three agents side by side.
- How we compare vendors
Why this family refuses to print an estimate as if it were a rate card.
- How to judge a vendor
Finish, pricing, and fit, including when another vendor is the better choice.
- What an AI support agent is
The unit of work behind a pricing conversation, defined without a vendor's rate.
- Helpdesk automation
Automation inside the desk you run, which is a different buy from a concierge.
- AI support glossary
Short definitions for the terms this pricing page assumes.
- Sainapse pricing
The plan fees, included tickets, and overage restated on this page.
- What an AI support layer is
A layer on the desk you keep. Not a rip-out, and not a chat window.
- End-to-end system write
The step a conversation price does not, by itself, include.
Price the work, not the slogan
Bring a month of tickets. We will show what a platform fee covers, and what it does not.