Interior Agent

An image- or camera-based system analyzes your interior space and recommends suitable furniture or design styles based on the room’s layout, colors, and existing items.

Comment

Inspiration

Finding furniture that truly fits a specific room is harder than it looks. Many interior apps show generic examples that don’t reflect a user’s actual space, layout, or existing furniture.
We were inspired to build Interior Agent to bridge this gap by directly understanding a real room from images or a camera and turning that understanding into practical, personalized interior recommendations.

What it does

Interior Agent analyzes a photo or live camera image of a room and understands its layout, style, and existing furniture.
Using Gemini’s advanced image understanding (Nano Banana) and search capabilities, it either:

  • Recommends new furniture that matches the user’s request, with direct purchase links, or
  • Keeps the existing furniture and suggests a new layout arrangement to better satisfy the user’s needs.

The system is designed to support monetization by recommending sponsored or advertised furniture links from partner brands.

How we built it

We built Interior Agent by combining:

  • Image and camera input for real-world room capture
  • Gemini’s Nano Banana image analysis to understand spatial structure, furniture, and style
  • Gemini search to find suitable furniture and design references
  • A recommendation layer that matches user intent with furniture products or layout changes

The output is a clear, actionable recommendation with visual reasoning and direct links.

Challenges we ran into

One major challenge was ensuring accurate understanding of room scale and furniture placement from a single image.
Another challenge was balancing personalization with monetization—recommending sponsored items without reducing user trust or design quality.

Accomplishments that we're proud of

  • Successfully analyzing real interiors from images, not templates
  • Generating recommendations that respect existing furniture instead of forcing replacements
  • Designing a realistic monetization model through ads and sponsorships without disrupting the user experience

What we learned

We learned that interior design recommendations become far more valuable when grounded in the user’s actual space.
We also learned that AI-driven visual understanding can go beyond aesthetics and support practical, real-world decision-making.

What's next for Interior Agent

Next, we plan to:

  • Add multi-angle image support for more accurate spatial reasoning
  • Enable interactive layout simulations
  • Improve personalization through user preferences and budget constraints
  • Expand partnerships with furniture brands for broader, high-quality recommendations

Built With

Share this project:

Updates