Research Microsite · refreshed quarterly

Case Study: Intercom's Fin AI Agent

51% conversation resolution at scale — what Fin actually does, what it doesn't.

TL;DR

51% conversation resolution at scale — what Fin actually does, what it doesn't. The argument is laid out below with direct quotes from the canonical primary sources — vendor docs, regulator publications, peer-reviewed papers — so the reasoning can be checked.

Summary

51% conversation resolution at scale — what Fin actually does, what it doesn't. The canonical material on this topic, with operator-relevant takeaways and outlinks for further reading.

Canonical sources

"Fin 2 has achieved an average 51% conversation resolution rate across Intercom's customer base."

What we read out of it

Three operator-grade takeaways:

  1. The state-of-art is moving fast enough that quarterly refresh is non-negotiable.
  2. Vendor-published claims dominate the public discourse; analyst publications add framing; peer-reviewed papers anchor the long-term picture.
  3. The field's real disagreements (skeptics' critiques) are genuine signal — not noise to filter out.

Sources cited

  1. Intercom — Fin AI Agent. https://www.intercom.com/fin — accessed 2026-05-07