Intercom Customers Reach 80 Percent AI Resolution With Fin and New Monitors Feature.

Rhidian Boobier from Numan and Kat Crichton from Glean joined Intercom's Ruth O'Brien to detail how their teams have scaled AI-first support operations without sacrificing quality. Both companies reported resolution rates exceeding 80 percent while maintaining or improving customer experience.

The discussion comes as Intercom recently launched Monitors, a feature that allows support teams to review 100 percent of AI-handled conversations. The tool addresses a key concern for companies deploying AI agents at scale: how to maintain oversight and build trust in automated systems.

"Teams are scaling to 100 percent involvement without sacrificing quality," according to the panel. The phrase suggests both companies have moved their entire support operations to an AI-first model, with Fin handling the majority of customer inquiries.

Monitors gives support leaders visibility into every conversation Fin handles, closing what has been a significant gap in AI agent deployment. Previously, teams relied on sampling or spot checks, which meant most AI interactions went unreviewed.

The feature allows teams to:

• Review 100 percent of AI-handled conversations, not just a sample
• Close feedback loops faster by identifying patterns in AI responses
• Build confidence in AI systems through systematic observation
• Catch edge cases and improve training data in real time

Support leaders can now track performance metrics across all conversations, rather than extrapolating from limited samples. This matters particularly as companies move from pilot programs to full deployment, where the volume of AI interactions can reach thousands per day.

Monitors represents a shift in how companies approach AI quality assurance. Instead of hoping AI performs well based on limited testing, teams can now verify performance continuously.

An 80 percent AI resolution rate means only 20 percent of customer inquiries require human intervention. For context, Intercom customer Solidcore reported saving 12,000 hours annually with Fin handling routine support tasks.

Both Glean and Numan operate in sectors where accuracy matters. Numan provides healthcare services, where incorrect information carries real risk. Glean offers enterprise search and knowledge management, where users expect precise answers. The fact that both companies feel comfortable running AI-first operations suggests the technology has crossed a reliability threshold.

The panel discussion did not detail which types of inquiries Fin handles versus those routed to human agents. Resolution rate alone does not indicate customer satisfaction, though the companies claimed both efficiency and experience improved.

Intercom has positioned Fin as a specialized AI agent rather than a general chatbot. The company recently launched specialized roles for Fin, starting with a sales agent variant, suggesting the platform is expanding beyond support.

The emphasis on monitoring and oversight reflects broader industry concerns about AI deployment. As companies like Crisp and others add AI capabilities to customer service tools, the question of how to verify AI performance becomes critical.

Monitors is available now to Intercom customers using Fin. The feature launches as companies face pressure to reduce support costs while maintaining quality, a balance that AI promises but does not always deliver without proper oversight.

The full customer panel discussion is available on Intercom's LinkedIn page.