Roomform AI

Inspiration

We started out with a simple 3d model generation pipeline, then got support from our college professors. Digging deeper, we saw there was a need for a shopping and design tool that was AI/3d-native, and not an image wrapper. We built Roomform to give people open choice, real products, and push creativity. Beyond a tool, we're hoping to build a community around democratizing design & retail.

What it does

It takes a scan of a space (using the iOS app) and turns it into a 3d replica on the web app. From there, users can add items by drag & dropping photos, shop in the marketplace with direct checkout links (has every web listing), and ask our AI for suggestions and rearrangements. Users can then render and post to the explore page, and collaborate using shared links.

How we built it

The main pipelines are hosted on a GPU cluster - no API.

Challenges we ran into

Months of iteration, failing, and trying again. Optimizing our pipelines to cut costs and generate 3d models quickly was the toughest hurdle.

Accomplishments that we're proud of

Successfully launched! Won the entrepreneurial prize at our college, where Roomform started.

What's next for Roomform AI

We're open to feedback from anyone! Roomform is being iterated on constantly - next features include: floor plan uploads, improving AI spatial intelligence, AI agent for retailers to bulk upload catalogs, in-store hybrid QR shopping system.

Built With

• django
• nextjs
• postgresql
• vercel