in-depth: Stop Touching Your Keyboard. Use This AI-Powered Microphone

London-based startup Relay is set to revolutionize human-computer interaction with the Relay Q, a dedicated AI-powered portable microphone designed to make typing obsolete. Co-founded by former Nothing employees Cookie Xu and Raymond Zhu, the company launched its macOS software today, with the innovative hardware scheduled for release in early 2027. This new device aims to offer a seamless, high-fidelity voice-to-text experience, leveraging advanced large language models to intelligently process spoken words into text, and even execute complex commands.

The surge in large language model capabilities has propelled voice-to-text technology forward, enabling tools to not just transcribe but also refine speech by removing filler words and clarifying unscripted thoughts. Despite these advancements, a significant hurdle remains: practical application in diverse environments. Raymond Zhu shared that during his time at Nothing, using transcription software for messages highlighted the challenge of finding an appropriate volume in an open office, leading to the vision for dedicated hardware.

Relay's initial offering is its macOS application, which can be activated in any text field using a customizable hotkey, such as the Fn key. Powered by Google's Gemini models, the software is designed to go beyond simple dictation. It intelligently cleans up spoken input, stripping out unnecessary words and refining ideas on the fly, and can even facilitate multilingual conversations with ease.

A core feature of the Relay app is its "Skills" functionality, which allows users to perform contextual actions directly through voice commands. This includes tasks like sending a message to a colleague in Slack or creating a calendar event in Google Calendar, although the beta version showed some inconsistencies. The software also offers extensive customization, letting users tailor punctuation habits for different contacts or automatically translate messages for international colleagues, ensuring appropriate tone and style for each interaction.

The forthcoming Relay Q hardware is envisioned as a cylindrical device with a microphone at one end, designed for portability and discreet use. Users can activate it by pressing the top while speaking, or clip a detachable button to their clothing for hands-free operation. Its specialized design promises to capture speech even when whispered, addressing the privacy concerns of speaking aloud in an office. The Relay Q will include a charging dock and plans for a MagSafe-compatible grip for mobile devices, aligning with a planned iOS and Android app launch.

While promising enhanced convenience, the Relay system necessitates significant system permissions, including access to the microphone and screen snapshots for its "Skills" feature. Relay states that user data is stored locally on the device and can be manually edited or deleted. When data is sent for processing, it travels through Relay’s servers en route to Google’s, but Relay itself does not retain the data. However, granting AI models "eyes" on an operating system inherently introduces privacy and security trade-offs, a concern previously highlighted with similar technologies like Microsoft Recall.

Relay enters a burgeoning market where AI-driven voice-to-text is rapidly evolving. Current competitors range from advanced software like Wispr Flow, which summarizes meetings, to enhanced on-device dictation in smartphones like Google's Pixel series. Other startups have also explored dedicated voice hardware, with products like Sandbar's Stream Ring and Pebble's Index 01 offering whispered dictation through wearable rings. Relay opted for a microphone form factor over a ring, seeking to differentiate itself in a crowded smart ring market.

The Relay Q hardware will be available in the first quarter of 2027 for $150, which includes a one-year subscription to the Relay software. After the initial year, the software subscription will cost $15 per month. It's important to note that the Relay software can be purchased and used independently of the Relay Q hardware.

Despite its innovative approach, the concept of widespread voice input in professional or personal settings raises questions. Some early reactions from users suggest skepticism about its practicality in environments like open-plan offices or family homes, where constant vocal input might create a noisy or disruptive atmosphere. The challenge remains for such technology to seamlessly integrate without altering social dynamics in shared spaces.

FAQ

Q: What is the Relay Q and when will it be available? A: The Relay Q is a dedicated, AI-powered portable microphone designed for high-fidelity voice-to-text interaction. Its accompanying macOS software is available now in beta, with the hardware itself expected to launch in early 2027.

Q: How does the Relay Q improve on existing voice-to-text solutions? A: Unlike typical software, the Relay Q is dedicated hardware designed to capture voice input accurately, even whispers, in various environments. It integrates with Google's Gemini models to intelligently clean up speech, perform contextual actions, and offer extensive customization for tone and language, addressing challenges like speaking volume in offices.

Q: What are the privacy implications of using Relay Q? A: Relay states that data is stored locally on the user's device and can be managed there. While data is processed through Relay's servers en route to Google's, Relay does not retain the data. However, the system requires broad permissions, including microphone access and screen snapshots for its advanced "Skills," which introduces inherent privacy and security considerations.