Overview
Fin is Intercom's AI-powered customer agent that handles customer support conversations from start to finish. It learns from your help center, past conversations, and internal docs to deliver accurate, human-quality responses. Fin handles service, sales, and e-commerce queries across chat, email, and social channels.
Our take
Fin resolves support conversations end to end, priced per resolution on top of Intercom, which is an unusually honest pricing model — you pay when it actually works. That alignment is the strongest thing about it. The number to interrogate is what counts as a resolution, since that definition determines both your bill and your quality measurement.
Worth it if
Intercom customers with documented help content and enough ticket volume to matter.
Look elsewhere if
Teams not on Intercom, or those without help documentation for it to ground answers in.
✨ Features
End-to-End Resolution
Resolves customer issues completely without human intervention.
Multi-Channel Support
Works across live chat, email, SMS, WhatsApp, and social media.
Knowledge Learning
Learns from help center articles, past conversations, and internal documentation.
Sales & E-commerce
Handles inbound sales questions and guides customers through purchase journeys.
⚖️ Pros & Cons
Pros
Cons
💰 Pricing
Essential
$29 /seat/mo
- Fin AI agent included
- Shared inbox and basic automation
- Pay per AI resolution ($0.99 each)
- Help center and live chat
Advanced
$85 /seat/mo
- Advanced AI and automation tools
- Multiple team inboxes
- Workflows and triggers
- Pay per AI resolution ($0.99 each)
Expert
$132 /seat/mo
- Collaboration and workload management
- Custom reporting and SLA rules
- Multi-brand support
- Pay per AI resolution ($0.99 each)
Categories
❓ Frequently Asked Questions
What counts as a resolution?
The definition is the whole commercial question and worth reading closely. A conversation that ends without escalation is not necessarily a customer whose problem was solved — sometimes they gave up. Track subsequent contact rate on resolved conversations, which distinguishes genuine resolution from deflection and is what you actually care about.
How much of my ticket volume can it handle?
For a business with decent documentation, commonly a quarter to a half of inbound, concentrated in repeat informational questions. Higher figures usually count deflection. The determining factor is your help content: the tool grounds answers in it, so poor documentation caps performance regardless of model quality.