Sesame's lifelike voice agents are on iPhone and Android now, ahead of glasses planned for 2027

At a glance

Best for: curious readers who want to try conversational voice AI on their own phone.

Skip it if: you need a hands-on verdict on the glasses; none exists yet.

Reviewers agree: Hollister, Patterson and Farouk found the voice natural.

Price: free on the US App Store, checked 14 October 2026; glasses unpriced.

Status: app in preview on iOS and Android; glasses planned for 2027.

The Verge's Sean Hollister wrote in February 2025 that Sesame's Maya was the first voice assistant he'd wanted to talk to more than once. In his hands-on with the public demo, he asked her to put herself into a fantasy adventure she was describing, and she did, with natural pauses and easy interruptions. PCWorld's Ben Patterson tried the iPhone app in June 2026 and sensed that Maya was listening and thinking as she talked, with no moment of feeling lectured.[1] [2]

Maya runs on Sesame's Conversational Speech Model. Its open-source CSM-1B release has one billion parameters and generates audio codes from text and audio, per TechCrunch's Kyle Wiggers in March 2025, and Sesame says a fine-tuned version powers Maya. The Verge reports a training set of about a million hours of public audio. Wiggers noted that Maya and Miles take breaths, hesitate and can be interrupted. In a June 2026 test, futurezone's Amir Farouk found it almost eerie how natural the agents sound, and on his first call he used up the app's 30-minute session limit.[3] [4] [1] [5]

The app, Sesame: Personal Agents, is listed as free on the US App Store for iPhones on iOS 18 or later (checked 14 October 2026); TechCrunch called the full experience free for the time being at its May 2026 launch. Patterson watched Maya run background searches mid-sentence and name a nearby restaurant without a long silence. Sesame says its 6 October preview release added a new voice model and agents with their own computer, in the app for iOS and Android. Farouk found no source links for answers and reports that a paywall is planned.[6] [7] [2] [3] [5]

The glasses are still a plan. Sesame says it's making all-day frames, designed together with the assistant, and its 6 October post says the collection will be made in Japan. The App Store listing and TechCrunch both give 2027 as the expected year, and Sesame promises more on the collection in the coming months. In February 2025 The Verge saw only a few small images of what look like early prototypes. As of 14 October 2026, no hands-on, price, specifications or camera and display details had surfaced.[3] [6] [7] [1]

Oculus co-founder and former CEO Brendan Iribe leads Sesame with Ankit Kumar, former CTO of Ubiquity6, per The Verge's February 2025 launch report. TechCrunch notes a $250 million Series B from Sequoia and others, raised before the app's beta opened. At launch The Verge listed Andreessen Horowitz, Spark Capital and Matrix Partners as backers, all big Oculus VR investors.[1] [7]

Patterson's praise came with a worry about where natural-sounding voice chat turns harmful. His own chat had questions but no probing, though colleague Mark Hachman found Maya uncomfortably inquisitive in February 2025. Wiggers reported that the open CSM-1B model has no real safeguards beyond an honor-system request, and on its Hugging Face demo he cloned his voice in under a minute. On privacy, the App Store listing promises no ads, no data sales and nothing saved in Incognito mode, while the privacy policy edited on 6 October 2026 says logged-in users' content may be used to train Sesame's models unless they opt out in account settings. Try the voice on your phone now, and judge the glasses when a price, specifications and microphone details appear.[2] [4] [6] [8]