Retail Innovation Prototype | A Smart Mirror

At NorthOut, we’re often challenged by our partners to come up with new innovation ideas, quickly prototype a solution, and create a plan to go full scale. Sometimes the conversations start with a focused problem statement like: ‘help me understand opportunities for machine learning for my supply chain operations’ while other begin more broadly: ‘I need a new digital engagement opportunity for my customers’ — either way, we quickly deploy an Innovation Squad and at lightspeed, get to work.

Our process is flexible and uniquely positioned for each challenge since we don’t believe in the cookie cutter consulting firm approach to project…technology is always evolving so why shouldn’t we?

That said, no matter what we do, we adhere to three clear standards:

  1. We start in the mind of our customers
    With close user involvement, products are more likely to meet users’ expectations and requirements. This leads to increased sales and lower costs incurred by customer services.
  2. We create a prototyped experience that is real – not just a powerpoint
    Creating a real experience further de-risks innovation exploration. Not only does building some real highlight any possible engineering challenges (more on that soon) it also provides stronger customer feedback validation while driving evangelism internally.
  3. We expect to be prepared to bring this to the greater organization
    Many organizations tackle a innovation exploration in a vacuum from stakeholders both on the business and technology side. This leads to stacks of old concepts gathering dusk on the shelf and a feeling that ‘innovation exploration’ never drives real ROI. At NorthOut, we certain to help partner both plan further development, creates business cases and prepare to drive full scale deployment.

Recently, a retailer ask us to imagine a digital experience not directly aligned with their brand of today to provide value to users in a high frequency environment. We love targeting ‘high frequency’ experiences – meaning places a user engages often – think buying coffee in the morning or brushing your teeth. After talking with a handful of users we realized there was a major opportunity in the bathroom, specifically with the bathroom mirror.

Given the brand’s ‘we here to help’ attitude and increasing cost effective hardware, we set our to explore the feasibility of creating a voice activated ‘smart mirror’ that while just a typical mirror on the surface, could provide a morning update, fashion insights and more on command. Much of this was inspired by the latest home exercising trend driven by Mirror Fitness.

We knew a few things were important to potential customers by talking to them:

  1. The mirror couldn’t be intrusive, or noticeably digital when off
  2. It’d have to be ‘always ready’ and low powered to not need replacement often
  3. It’d have to be internet connected and personalized

On the partner side, their wishlist included:

  1. Able to be extendable over the air and configurable
  2. Exploration would have to be done with a limited, discretionary budget
  3. Understand how the architecture approach could enable the ability to integrate with their purchase

So we got to work!

Over a period of 3 weeks, we went from user interviews to building a prototype to details future roadmap documentation and enables a demo for wider organization stakeholders. The beauty of the mirror itself was held secondary (eventually we’ll build a pretty frame or build truly custom) and we had a bit of trouble getting through TSA but minus a few scratches we were able to demo to a wide range of stakeholders.

From the hardware side: We started by investigating other experiences and realized that we could combine a 2 way mirror, modified OLED monitor and a Raspberry Pi (with microphone expansion) to match what a custom hardware build might be someday. This was all doable under $500.

For software, we wrote an Electron application with a simple Node backend that integrated with the New York Times, Local Weather, Google Maps and a mocked API that matched the retailer’s specifications. For voice recognition, we used the TensorFlow Raspberry PI SDK and train on variants of ‘Mirror Mirror on the Wall’.


Alongside detailed documentation, our demo drove excitement across teams to continue investing in exploring what the experience could be and started wider conversation about the brand’s future approach to digital. We’re currently working on a much larger prototype to display in a real on-location bathroom.

Want to explore how frontier technology could drive your brand? Contact NorthOut.


Steve Jobs