Fin

The AI customer service agent built by Intercom

An AI agent that resolves customer support conversations end to end across chat, email and voice, priced per resolution rather than per seat, and deployable on top of most major helpdesks.

Screenshot of the Fin homepage

Fin is the AI customer service agent built by Intercom, designed to hold and resolve full support conversations rather than just suggesting replies for a human to send. It reads a knowledge base and a set of team-defined “procedures,” then handles the conversation end to end — answering, asking clarifying questions, or handing off to a person when it’s out of its depth.

The pitch that sets it apart from most support bots is the pricing model: instead of a flat license, Intercom bills per resolved outcome, so a quiet month costs less than a busy one. That suits support teams with spiky or seasonal volume, and it has pushed Fin to expand beyond Intercom’s own helpdesk to sit on top of Salesforce, Freshdesk and other platforms — a sign the product is being positioned as a standalone AI agent layer rather than an Intercom-only feature.

It’s a stronger fit for teams that already have a well-organized knowledge base to train it on; garbage-in support docs produce mediocre resolutions no matter how good the underlying model is. And because pricing is usage-based and quoted per account, it’s genuinely hard to budget for until you have real conversation volume flowing through it — worth piloting on a slice of ticket volume before committing broadly.

Features

Outcome-based resolutions

Fin only bills for conversations it actually resolves, hands off, or disqualifies — not for every message it touches.

Procedures

Support teams encode their own workflows and policies as step-by-step procedures instead of relying on generic training data.

Multi-channel coverage

Runs across chat, email, voice and Slack from a single configuration rather than a separate bot per channel.

Runs on your existing helpdesk

Deploys over Zendesk, Salesforce and others as well as Intercom, so adopting it is not also a migration.

Knowledge from content you already have

Ingests the help centre, past tickets, PDFs and public pages rather than requiring a separate intent tree to be authored.

Resolution reporting

Reports on questions actually answered rather than on deflection, which is the number the outcome-based pricing is charged against.

Use cases

  • Deflecting high-volume, repetitive support tickets automatically
  • Handling first-line triage before handing complex cases to a human agent
  • Running AI-driven voice support without a separate IVR system
  • Adding an AI agent on top of Zendesk or Salesforce without replacing it
  • Measuring real resolution rate rather than guessing at deflection
  • Covering out-of-hours support without staffing a night shift

Compare Fin head to head

Side-by-side comparisons, on pricing, platforms and where each one wins.