Spatial Sparky

Generative AI in Retail · Student Capstone

What If Shopping Started With a Photo?

Spatial Sparky reimagines product discovery as a visual planning experience: upload your space, describe the moment, and get recommendations that actually fit.

Student Team: Sean Tsang and Omar Bawa
Faculty Advisor: Prof. Kirthi Kalyanam
Program: Generative AI in Retail
Cohort: Spring 2026

Sean Tsang and Omar Bawa presenting the Spatial Sparky capstone project in a classroom

Most shopping does not begin with a perfect search query. It usually begins with a real-life moment: a dorm room that feels empty, an apartment that needs to be ready for friends, a backyard birthday coming up, or a student event that needs chairs, lights, decorations, tableware, games, and cleanup supplies.

Then the tabs start opening. One tab becomes five. Five becomes twenty. Soon, “What should I buy?” turns into: Will this fit? Will it match? Is it within budget? Am I forgetting something?

Before

chairs lights plates decor budget cleanup games snacks

A shopper compares items across tabs and tries to imagine whether everything will work together.

After

1 photo
+ context

Spatial Sparky turns the space, occasion, guest count, and budget into one smarter shopping plan.

The Gap We Noticed

Shopping assistants can hear the request—but not always see the space.

Conversational shopping makes product discovery feel more natural because customers can describe what they need instead of relying only on keywords.

But one layer is often missing: the physical environment where those products will be used. Without that context, it is difficult to judge scale, color, layout, lighting, or whether several items will work together.

Meet Spatial Sparky

A “shop my space” experience

Users upload a photo of a dorm room, apartment, backyard, patio, living room, or event area, then add details such as event type, theme, guest count, date, setting, and budget.

Spatial Sparky analyzes layout, lighting, color palette, open areas, and existing furniture, then recommends products that make sense for that specific space and occasion.

Instead of only asking, “What are you looking for?” Spatial Sparky asks: “What will actually work here?”

1 Upload a photo Start with the real space.
2 Add details Include the theme, guests, date, and budget.
3 Analyze context Review layout, lighting, colors, and furniture.
4 Recommend Surface items that fit the space and moment.

One Real-Life Scenario

A backyard birthday, planned from one image

Imagine uploading a yard photo, choosing a warm outdoor theme, setting a $150 budget, and planning for 12 guests.

12 guests $150 budget Outdoor evening

Seating Foldable chairs that match the open space.

Lighting String lights or lanterns for the mood.

Tableware Plates, cups, napkins, and utensils.

Cleanup Trash bags and wipes people often forget.

The Larger Vision

From product search to shopping confidence

Building Spatial Sparky helped us think about retail AI beyond speed. The future of shopping is not only about finding products faster. It is about helping people make better decisions in the context of their real lives.

The prototype begins with one retail catalog, but the larger vision is cross-retailer: platforms working around the customer’s space, budget, and occasion rather than treating every product search in isolation.