Fin, the AI agent formerly sold as Intercom, charges $0.99 per outcome plus an Intercom helpdesk seat, which starts at $29 a month on annual billing, or $39 billed monthly (checked October 2026). Your monthly bill is roughly the number of outcomes multiplied by $0.99, plus your seats. For one seat, and assuming every conversation the AI handles counts as an outcome, that works out to about $128 a month at 100 conversations, $326 at 300 and $1,019 at 1,000.
The rest of this post shows that arithmetic step by step, explains what counts as an outcome, and covers what moves the number up or down. Prices change, especially after an acquisition, so check Intercom's own pricing page before you budget.
What Fin charges for
As of our October 2026 check, the pricing has two parts.
- Outcomes at $0.99 each. You pay for results the AI delivers, not for every message.
- Helpdesk seats for the people on your team, at $29, $85 or $132 per seat per month on annual billing ($39, $99 or $139 billed monthly), depending on the tier.
There is a 14-day trial rather than a free plan. If you use Fin on its own alongside a different helpdesk, there was a minimum of about 50 outcomes when we checked.
Ownership matters here too. Salesforce completed its acquisition of Fin in September 2026 and has said it plans to fold Fin into Agentforce. Plans and pricing may change as that happens, so re-check the figures rather than rely on any article, this one included.
What counts as an outcome
This definition drives the whole bill, so it is worth reading slowly. By Intercom's definition, an outcome is counted when Fin resolves a conversation. That happens either when the customer confirms the answer helped, or when they do not ask for further help. Certain handoffs to your team that happen as part of a procedure Fin runs also count.
The second condition is the one to notice. A visitor who reads the answer and closes the tab has not asked for further help. In practice that means most conversations the AI handles to the end can count, not only the ones where a customer clicks a thumbs up.
For budgeting, the safest assumption is that every AI-handled conversation counts as an outcome. That gives you a ceiling. Your real bill will be at or below it.
Worked examples at 100, 300 and 1,000 conversations
Assumptions for all three: one seat at the lowest annual tier of $29 a month, every AI-handled conversation counts as one outcome, and no other add-ons.
| AI-handled conversations a month | Outcome cost | Plus one seat | Monthly total | Yearly total |
|---|---|---|---|---|
| 100 | 100 x $0.99 = $99.00 | $29.00 | $128.00 | $1,536 |
| 300 | 300 x $0.99 = $297.00 | $29.00 | $326.00 | $3,912 |
| 1,000 | 1,000 x $0.99 = $990.00 | $29.00 | $1,019.00 | $12,228 |
Two adjustments are worth running for your own case.
More seats. Each extra person at the entry tier adds $29 a month. On the middle tier at $85 per seat, two seats cost $170, so 300 conversations would come to $297 + $170 = $467 a month.
Fewer outcomes. If you expect some conversations to end without an outcome, reduce the count. As a hypothetical, if 210 of 300 conversations counted, the outcome cost would be 210 x $0.99 = $207.90, and the total with one seat $226.90.
For reference, Intercom's own calculator example puts 120 conversations at $109.09 a month. That is below the every-conversation arithmetic, which would give 120 x $0.99 + $29 = $147.80, so treat our figures as the ceiling for each volume rather than a forecast.
What pushes the bill up
Growth. The outcome line scales in a straight line. Double your traffic and that part of the bill doubles.
Busy months. A seasonal peak or a product problem that sends everyone to chat lands on that month's invoice. Budget for your busiest month, not your average one.
Better answers. This is the odd part of per-outcome pricing. The more content you give the AI and the better it gets, the more conversations it resolves, and the more you pay. Improving your help content is good for customers and still increases the bill.
None of this makes per-outcome pricing wrong. You pay for work done, and for a support team already running Intercom's helpdesk, adding Fin can make sense. It just means the cost is a variable, and you should forecast it like one. Our guide to how AI chatbot pricing models work compares per-outcome, per-message and flat pricing side by side.
How to estimate your own number
- Count chat or email conversations per month for the last three months, and note the highest.
- Estimate the share the AI would handle to the end. If unsure, use all of them.
- Multiply by $0.99.
- Add your seats at the tier you need.
- Repeat with your busiest month to see the peak.
The flat-price contrast
A flat-priced website chatbot works the other way: a fixed monthly fee with an allowance, and no charge per conversation. SpideyChat's allowance is measured in AI tokens, which works out to about 375 replies a month on Free, about 750 on Starter at $9.99 and about 3,750 on Pro at $29.99.
A conversation usually involves several replies. Assuming three replies per conversation:
| Conversations a month | Replies (assumed 3 each) | SpideyChat plan that covers it | Monthly price |
|---|---|---|---|
| 100 | about 300 | Starter (or Free, with little headroom) | $9.99 |
| 300 | about 900 | Pro | $29.99 |
| 1,000 | about 3,000 | Pro | $29.99 |
The trade-off is a ceiling. If an account uses its month's allowance, the chat pauses replies until the next month or an upgrade, instead of billing more. It is also a different product: SpideyChat is a chatbot for your website that answers from your content, captures leads, opens tickets and hands over to a person when someone is online. It is not a full helpdesk with a shared inbox and help centre. Our SpideyChat vs Intercom comparison lays out where each fits, and the Intercom Fin alternatives list compares other options with prices checked in October 2026.
If a flat fee suits your volume, the pricing page has the full plan details, and you can try it on your own website without signing up to see how it answers before comparing costs.