Sesame launches its iPhone voice-AI app in 39 countries

On 28 May, Sesame released a public preview of its iPhone voice app in 39 countries with four named AI agents (Maya, Miles, Simone, Charlie). It runs on the company's Conversational Speech Model (CSM) — an AI voice system engineered for natural-sounding prosody: the rhythm, pauses, and tone that make speech feel human rather than synthesised. To keep conversation flowing, it can run searches while speaking instead of freezing to think. Sesame previously open-sourced its CSM components, including the 1B 'CSM-1B' backbone, under Apache 2.0. The app is free for now (a short waitlist is possible); an Android preview is promised but undated. Co-founded by Oculus's Brendan Iribe and Ankit Kumar, Sesame raised a $250M Sequoia-led round before launch.

Voice is becoming the next AI assistant battleground. Apple overhauled Siri AI at WWDC 2026. OpenAI has a realtime voice API. Sesame enters as the dedicated, voice-only bet — smaller than the incumbents, but focused entirely on the one dimension where it has the strongest claim. The tech is converging; the win goes to whoever makes talking-to-AI a daily habit. Sesame's whole wager is that natural-enough prosody finally tips that calculation — that when the AI sounds human, the awkwardness of talking to your phone in public finally drops.

What's next. Watch for: (1) any retention data Sesame or third-party analytics firms surface in the weeks after launch; (2) an Android or web announcement; (3) a monetisation reveal — the free preview won't run forever. If retention holds, this is an important story. If it doesn't, add it to a long and distinguished list.