Conversational AI: how to make users want to talk to our product?

Conversational AI is transforming how people interact with our products. It's the famous shift from "browsing" to "asking": users no longer navigate, they formulate an intention and expect an immediate and contextualized response (via chat, voice, or even video). In this new paradigm, how do we create a product that really makes people want to talk? And how do we make that conversation useful and engaging? I will share two concrete cases of AI-native products I designed as Founding Designer at Hexa, the startup studio behind 50 startups including Aircall, Spendesk, and Front: Rose, a marketing agent for B2B sites where the challenge was to move beyond the image of "useless chat support" to create real business value and an interaction that engages and converts. Verso, an audio/video interview agent where the challenge was to establish enough trust and fluidity to obtain qualitative insights... with an AI opposite you. For each, I will detail the design choices we made over the iterations, and how these micro-interactions directly improved engagement and the quality of exchanges: animation, follow-up questions, reformulations... The approach is deliberately focused on practice, illustrated with real examples and field results. The talk is mainly aimed at designers and PMs who want to understand how to design more engaging conversational AI interactions. This is a major turning point in how we think about interaction design.