UX case study · Voice-first AI
EPOC
An AI system that manages IoT devices and data inputs to deliver real-time, actionable knowledge
- UX problems framed: trust, information fatigue, privacy, syntax, health
- 5
- concept project, from research to hi-fi prototype
- 2017
- opportunity areas found in the market
- 3
- concepts explored: Huic and Epoc
- 2
Information doubles every year, and no one can take full advantage of it. Epoc is an AI that manages a person's data inputs and IoT devices so they don't have to, filtering out what's irrelevant and delivering close to real-time, actionable knowledge.
The idea
E-consciousness is the external awareness of AI as an extension of a person's being. Today we are dependent on, and occupied by, our hands and eyes. Even with that concentrated attention, we remain unaware of much of the information available.
The Epoc family of products manages a user's IoT devices and mental trends, so the right information reaches them at the right moment.
The Epoc prototype.
The product family: one hub, a mobile app, and smaller connected devices. 3D model by Noel Gonzalez.
The problem
Five UX problems needed solving: trust, information fatigue, privacy, syntax, and health. The common thread is a lack of accurate, clear information reaching the user in real time, so the design prioritizes trust and privacy.
We framed the information environment as three things that have to line up: the context a person is in, the content that matters to them, and the user, who is multitasking and always looking for more accurate information.
The proposed answer: an ear wearable for hands-free voice, and a mobile app for a customized interface.
Research
I led the research. Secondary research traced voice assistants from IBM's Shoebox in 1961 through Siri, Cortana, Alexa and Google Home to Apple's HomePod in 2017.
For primary research, a survey asked people what they think about during the drive home from work, what they stop for, and what information they want before they arrive.
The team also drew up personas to represent the people we heard from.
A user journey map then followed one person from 60 minutes before the end of the workday to bedtime. The pain points and opportunities are marked in red.
Finding the gap
We plotted today's assistants and smart speakers on a psycho-aesthetic map, then averaged two individual maps into one. Three opportunity areas stood out in the current market.
Two concepts
The opportunity areas became two concepts. Huic, named for a huntsman's call to cheer on hounds, is a shopping companion that takes center stage in a store.
Storyboard sketches here and below by Rithesh Rama.
Epoc sits in a different area of the map. It supports the user across the day, from shopping to the drive home.
Both concepts were then laid against the user's own journey.
Wireframes and testing
Low-fidelity wireframes of the mobile app came next.
A quasi-empirical UX evaluation flagged the hierarchy, the drop-down list, the slide-and-delete interactions, search, linking with other apps such as Spotify, Pandora and Uber Eats, and repeated information architecture.
The Epoc network
Epoc has to work across many devices, so we mapped its network on a whiteboard.
The map became the user-Epoc network: the devices, services and people around a user, from streaming and social apps to the smart home.
Overlaid on a home floor plan, the network shows where Epoc devices sit and the zones they cover.
High-fidelity design
Epoc AI manages all data inputs so the user won't have to, giving the user close to real-time, actionable knowledge.
Testing in the drive
The prototype was tested for in-drive assistance, giving real-time information on the topic at hand and helping the user plan and carry out what they were gathering.
Participants said:
- "Faster felt real"
- "More spontaneous feedback"
- "Sometime you really need advice"
- "Not invasive"
- "Epoc should ask for forgiveness rather than permission"
- "It kind of feels like a third wheel"
- "Match user interest"