Designing Intercom's AI voice agent

Overview

Phone is still a dominant customer support channel, accounting +30% of the entire support market. With a growing market demand for AI Voicebots, Intercom explored a new opportunity: an AI agent for phone support. Unlike chat or text, voice introduces a high bar for latency, clarity, and human-like interaction which required a ground-up rethink of how AI communicates.

This was a 0→1 initiative, and the product needed to feel natural, human-like and trustworthy on a live call.

Role

Staff Product Designer

• Led the design of both end-user voice interactions and internal tools to deploy a Voice AI agent on a live phone line,
• Key design areas included defining voice conversation flows, fallback strategies, and seamless agent-to-human handoffs,
• Collaborated closely within a small, cross-functional team consisting of a product manager, AI engineers, and machine learning scientists.

Outcomes

• Launched Fin over Phone into early beta, with early customers adopting and deploying it to end-users,
• Integrated the Voice AI agent into existing phone functionality, including callbacks and out-of-office call handling,
• Delivered the Voice Playground, an interactive environment where customers could test, refine, and better understand their AI voice agents and deployment process.