Building a voice interface for Headspace

Headspace’s voice apps now have over 1 million subscribers. Speaking at Adobe MAX last week, Head of Product Randhir Vieira explained how the mindfulness brand worked with Rain to bring meditation and sleep exercises to Google Assistant and Amazon Alexa

Founded by former Buddhist monk Andy Puddicombe and ex ad man Rich Pierson, Headspace now has almost 30 million users. The brand’s app delivers short mindfulness exercises voiced by Puddicombe along with animations created by London studio Nexus. With the help of colourful artwork and Puddicombe’s soothing voice, these exercises have helped users carve out time for focus and relaxation each day.

For the past two years, Headspace has also been developing an experience for voice-activated devices. Sleep by Headspace can be accessed through Google Assistant and Amazon Alexa as well as the main Headspace app, and provides ambient sounds and guided meditation exercises to help users drift off. Users can choose from a range of ‘Wind Downs’, ‘Sleep Sounds’ or ‘Sleepcasts’ (ambient tracks which combine soothing vocals and visualisation techniques to help them relax). Current Sleepcasts include Midnight Launderette, Beachcomber and Desert Campfire.

Headspace worked with Rain – an agency that specialises in developing voice strategies for clients – to create Sleep and determine how voice could fit into the brand’s wider goal of bringing mindfulness to the masses. The brand launched Sleep on Amazon Alexa in 2017 and has since launched it on Google Assistant. Speaking at Adobe MAX in Los Angeles last week, Head of Product Randhir Vieira said its voice apps now have over 1 million subscribers.

Explaining the company’s decision to invest in VUI (voice user interfaces), Vieira said: “Looking back, it seems obvious, but around two years, ago we had to make a decision on whether we wanted to invest in voice platforms or not. One of the graphics we saw was the adoption curve for voice enabled device growth in the US [a report compiled in 2017 estimates that 55% of households in the US will have voice-activated smart speakers by 2022], so our first question was, can we afford not to? And our next question was, what were people doing [on their voice activated devices]? It wasn’t obvious to us that we were a good fit for the way people were using AI and voice devices in general, so it was a toss up between these high rates of adoption and not seeing a clear fit [in terms of] how people were using them,” he said.