The Shelf Elf: visual merchandising intelligence

Inspiration

I visited a wonderful local Asian supermarket here in London and realised I had no idea what was on the shelves. I knew there would be some really interesting products, but had no clue where to start. My wife said, 'Surely there's an AI app for this?' and I realised that was the problem I could solve.

What it does

Shelf Elf is a two-sided application. Consumers get a really useful AI assistant to help them understand what's on the shelves in front of them. Simply video or snap a picture of the products on the shelves you are interested in. The app automatically uses the power of Gemini 3 visual capabilities to identify products. The user then sees relevant info about the products in front of them. Translation is built in, so great for travellers too. Brands and Retailers leverage the app as well. Not only does it free up store staff time helping customers identify products they want to buy, in addition we're recapturing highly valuable merchandising data. The images and videos captured by 'crowd-sourced' consumers provide a wealth of data for stores and brands to understand the reality of their shelf stocking and merchandising. Gemini 3 visual image understanding assesses the shelf stocking, product placement, and visual merchandising, and provides detailed scoring and analysis. It also makes actionable recommendations for improvements. Individual store managers can get near real-time feedback on the state of their store. Aggregating up to regional and national levels is extremely powerful. By leveraging the power of the 'crowd' of consumers The Shelf Elf helps retailers and brands optimise their in-store merchandising on a previously unimaginable scale.

How we built it

In less than two days entirely using AI Studio.

Challenges we ran into

Very few issues. Remarkably, given the complexity of the app, the trickiest thing was finding a way to host the logo image.

Accomplishments that we're proud of

I'm really pleased at the quality and sophistication of the app that I built with AI Studio. I've only used AI Studio for a few weeks. To go from Idea to Done in two days is really extraordinary.

What we learned

The main thing I've learned is to be creative and ambitious. Agentic coding in general has come so far in just a year. I also learned that poorly stocked shelves is a $125 billion dollar problem.

What's next for The Shelf Elf?

We've got a good MVP to illustrate the key capabilities. The consumer facing version is useful as it stands today. I'd like to make that fully featured with auth and database storage. As a platform that uniquely bridges the digital/in-store divide, The Shelf Elf drives high levels of engagement. An AI chatbot to give product advice eases friction and drives conversion in-store. There is huge scope for creativity, such as games, e.g. sending consumers to find a promotional item in store. Further expansion would be around the global commercial potential of the B2B-side, delivering a brand-new type of data to merchandisers and brands. Creating a multi-tenant application for global FMCG and retailers to analyse their real-world performance digitally. Poor visual merchandising costs 3% of the total market. Being able to solve a $125 billion dollar problem is pretty exciting.

Built With

• aistudio
• localstorage
• react