The Intercom alternative
for teams who stopped counting seats.

01

Fin's meter is better than ours, and here is exactly how

The first section argues for the competitor, because on this axis they are right.

Intercom bills $0.99 per outcome, and the list of things that are not an outcome is long. Verbatim: "If a customer asks for a human or shows frustration, Fin escalates based on its default logic. You are not billed for these escalations." A Procedure that hits a technical error is not billed. A conversation "simply passed to your team without an outcome" is not billed. A clarifying question the customer never answers auto-closes and "is not a billable outcome."

It even goes backwards. If a customer returns to a conversation Fin had resolved and asks for more help, "that resolution will be deducted and not charged." Very few meters in software refund themselves.

Outlearn bills every AI reply, the one that missed included. Their design is better. Ours is attached to a smaller total bill.

02

What an outcome actually includes

The word changed, and the word is doing work.

Intercom used to say resolution. They now say outcome, and an outcome is wider. Their published list has four kinds: a resolution, a Procedure handoff, a disqualification, and a qualification.

A Procedure handoff is billable. Verbatim: "A Procedure handoff is an outcome where Fin successfully executes a Procedure that you've configured to end in a handoff to a human or a workflow." So a general escalation is free, but a handoff you deliberately designed costs $0.99. Defensible, since Fin did the work you asked. Also not what people picture when they read "pay only for resolutions".

A qualification costs $9.99. Ten times the standard rate, for Fin matching a prospect to criteria you defined. If you point Fin at sales traffic, that is the line to model.

And a resolution does not need anyone to confirm it. Intercom counts a confirmed resolution when the customer says something like "that helped", and an assumed resolution when they simply stop replying. The assumed one fires after 24 hours of inactivity at the same $0.99. Fin has to have given a real answer first, which is a genuine safeguard, but silence still bills.

One more if you buy Fin standalone. There is no base plan: fin.ai/pricing sets a "50 outcomes per month minimum" at $0.99 each, so the floor is what fifty outcomes cost whether or not you use them. A usage limit below that does not reduce it: "if your subscription includes 50 outcomes, setting a limit of 30 won't reduce your charge, you'll still be billed for 50."

03

The bill Fin sits on, and what Fin is allowed to read

Fin is included on every Intercom plan, which is really just where the seat charge lives.

Annual seats are $29 on Essential, $85 on Advanced and $132 on Expert. Their pricing page's own structured data carries the monthly price for all three: $39, $99 and $139. What gets expensive is what the tiers gate. SAML, HIPAA support, SLAs and multibrand are on Expert only, so a security review asking for SSO does not move you one tier, it moves you to $132 a seat. Then Copilot is $29 per agent annually, Pro from $99, Proactive Support Plus $99, and WhatsApp, SMS, bulk email and phone are separately metered with no published rates. Ten seats on Advanced is $10,200 a year before one Fin outcome. Our seat line is zero. Be fair though: at one or two seats on Essential, Intercom is $29 a seat and we are not the cheap option.

Now the knowledge, and this is a published list rather than a guess. Fin reads website crawls, Zendesk, Notion, Confluence, Guru and Box, plus PDF and DOCX uploads, Salesforce Knowledge and Freshdesk, plus Intercom's own Articles, Snippets and Webpages. A more serious list than most competitors publish.

Two things are missing from it: SharePoint and Google Drive. For a company running Microsoft 365 or Workspace that is not a footnote, because that is where the answers actually are. Outlearn reads both, plus OneDrive, Dropbox, Box, GitHub, Slack, Teams, Gmail and Google Chat.

The Confluence support has edges too. Cloud only, "not on-premises", and "Only 'global' type Confluence spaces are imported", so a team space does not come across. Uploads cap at 100 per workspace. And a synced Zendesk help centre has to be fully published or Fin gets none of it.

The freshness difference is not one number. Intercom publishes a different cadence per source, from hourly for Zendesk articles down to "synced weekly" for your own webpages, and 14 days for a site heavy enough to be expensive to crawl. So if your answers live in Zendesk, Fin is fresher than a day; if they live on your own website, it is a week. Outlearn re-reads all twenty systems every 30 minutes.

Synced 20 mins ago, against Intercom's 24-hour re-sync. To its right, 4 failed in red.

04

What Outlearn does instead, and what it cannot do

It answers with the receipt attached. A 3 Sources chip under each answer, so anyone can open the articles it read. Pass a signed token and retrieval scopes to whoever is asking: public docs for a visitor, their plan's articles for a customer, internal runbooks for an employee. The limit you will find in week two: scoping works on your knowledge base and not on cloud-file connectors, which do not import users. It also runs HTTP actions against your own endpoints, as Fin does.

Now the gaps, and they are real. No shared inbox, no queue, no Messenger, no outbound campaigns, no product tours. No voice product against Fin Voice. And Fin's meter really is better than ours: if your volume is spiky and your deflection rate is low, a meter that charges nothing for failure is worth real money, and that is theirs.

Step 1 is your endpoint and your auth headers. Step 2 is a sentence, not a rules tree: "Describe when Outlearn should run this action."