[B18] GoodDealsAI

Inspiration

We are students living in Paris, and as you probably know, groceries are expensive there. In addition, the apps and services we tried to compare shops and products are not really efficient and lack information.

It’s been a while since we discussed the business opportunity to create and deploy an efficient product for a frequent and boring task, such as grocery shopping.

What it does

GoodDealsAI is a software developed for Meta Quest S3 and designed to run on Meta Glasses. While worn, it uses camera access to scan supermarket aisles, detecting items from your grocery list. It then identifies price tags and suggests the best deals based on your preferences, such as dietary restrictions or budget constraints.

How we built it

Using Unity 6, the project is based on a YOLO nano infering model that detects price tags and place them in the world. Meanwhile an OpenAI requests analyze the view to detect groceries from our list and returns a bounding box.

With those, we filter the price tags and isolate only what we need.

At the end we use the informations we gathered to suggest the best product.

Challenges we ran into

TIME

And a lot of weird bug on MetaSDK, still not solved to this day but we found workarounds.

Also the concept include the dependency of being at a real supermarket. The systems used change a lot between our 3D models tests and reallife seethrough.

Accomplishments that we're proud of

Everything done there, we had astonishing results at some point and it showed us that our goal is reachable.

Secondly making the trip to Barcelona enhanced the experience. It was a great idea.

What we learned

First of all, we learned many things about Barcelona, which was our first time there!

Technically, we learned so much. Initially, we are junior Gameplay Programmers with little XR and AI background. Everything had to be learnt, felt like we dived in deep water.

What’s next for [B18] GoodDeedsA

We built a functional core in three days with AR tracking, basic price comparisons, and gesture controls. But the real goal is a smarter, data driven app that learns from user habits, preferences, and routines to deliver personalized product suggestions.

To make this happen, we need better product analysis, tracking nutrients, brand popularity, discounts, and trends, so recommendations feel tailored, not random.

The challenge is time and resources. As students, we can’t build everything at once. The next step is to strengthen the core while finding ways to scale smarter, through partnerships, funding, or better tools.

The prototype is done. Now, we make it truly intelligent.

Built With

• c#
• openai
• unity
• yolo