Inspiration

People buy products every day but rarely understand what’s actually inside the box, behind the label, or whether what they received matches what was promised. We wanted to build an AI assistant that helps users understand products better and make informed decisions, without fear-based or authoritative judgments.

What it does

Scan My Product helps users understand products using images, links, and additional evidence. Users can: a. Upload a product image or link b. Get a clear explanation of what the product appears to be c. Understand ingredients, materials, or key features d. Receive neutral guidance on how to verify product consistency and quality e. Add extra information (labels, receipts, details) for deeper review The system focuses on explanation and guidance, not final judgments.

How we built it

We used Gemini 3 for multimodal reasoning across images, text, and web data. The system works in stages: a. Initial analysis using visual and contextual understanding b. Follow-up analysis when users provide additional evidence c. Guided verification that suggests practical next steps instead of hard conclusions

This staged approach keeps outputs accurate, cautious, and user-centric.

Challenges we ran into

Designing AI outputs that are informative without being authoritative a. Handling inconsistent labels, images, and user-provided data b. Avoiding definitive language while still delivering useful insights c. Balancing safety, clarity, and real-world usefulness

Accomplishments that we're proud of

Built a multi-stage product understanding flow using Gemini Created a safe, non-judgmental verification approach Successfully analyzed diverse product categories (food, apparel, consumer goods) Designed prompts that prioritize user empowerment over AI decisions

What we learned

AI is most effective when it supports human decision-making, not replaces it Breaking analysis into stages improves accuracy and trust Language choice matters as much as technical capability Responsible AI design leads to better real-world usability

What's next for Scan My Product

a. Smarter follow-up analysis using more evidence types b. Better UX for guided verification flows c. Expansion into sustainability, safety, and quality insights d. expert and brand-reference integrations. [Optional]

Built With

Share this project:

Updates