Project type

Background

ThirdEye Labs was a London based machine learning startup acquired by Standard Cognition in 2021. I was the first and only product designer at ThirdEye between 2019-2021, working with the machine learning and software engineering teams on translating machine learning outcomes to end users. ThirdEye created a computer vision model to detect concealments in retail environments in real time, creating image and video alerts of concealments, potential thefts, occurring in the store. These alerts would be sent to the end-user, primarily a security guard, with the metadata of where and when the alert was created.

My role

User research into security work in retail, understanding existing workflows and challenges

Designing a mobile app for users to view and action AI alerts on the go, including:

User journeys
Wireframing
UI
Usability testing
QA process
Continuous performance research

Challenges

Experience

Making decisions based on AI alerts without knowing the reasoning behind an alert - why was this activity picked up? How can I know if the model was wrong?
For most of the users this was their first interaction with machine learning outside voice assistants.
The user expectations for machine learning were very high.
Incorporating ThirdEye alerts as a new tool into existing workflows and processes.

Technology

All end users were using the app on the same device, a simple Android smartphone, with limited storage, slow operating speed and, by today’s standards, small screens.
The devices didn’t have a mobile data connection and relied on often intermittent in-store Wi-Fi

Successes

User success

Business

Users learned how to use the app and alerts incredibly quickly, most of the time learning from their colleagues on the job.
Positive feedback: users described the app as their “second set of eyes behind their heads”
Used in over 50 supermarkets in the UK and Europe, by hundreds of security guards