Inspiration

Shopping on Amazon today feels like doing data analysis. Endless reviews and conflicting opinions cause severe decision paralysis. We built BuyWise to cut through the noise, giving users their time and confidence back by turning chaotic text into instant, actionable insights.

What it does

BuyWise is a local-first Chrome extension that acts as your personal shopping copilot.

Instant Summaries: Extracts pros, cons, and verdicts directly from product pages.
Zero-Hallucination Q&A: Answers questions strictly based on real buyer reviews. If real buyers didn't say it, our AI won't invent it.
Product PK (The "AI Arena"): Select two competing items, and our AI runs a deep, side-by-side comparison to declare a winner.

How we built it

Frontend: A Manifest V3 Chrome extension keeping user data completely private using chrome.storage.local.
Backend: A fast and robust API built with Python and FastAPI.
AI Engine: We feed scraped page data into the ASI1 model. We chose ASI1 for its fast inference and strong performance on structured reasoning tasks.

Challenges we ran into

Our biggest hurdle was stopping AI hallucinations. In e-commerce, a made-up spec ruins trust instantly. We spent hours iterating on prompt engineering to force the model to answer strictly based on the provided review context. Additionally, managing asynchronous state smoothly between the extension's content scripts, popup, and background worker was a tough engineering puzzle.

Accomplishments that we're proud of

We are incredibly proud of our Product PK engine. Watching our backend take unstructured, messy reviews from two completely different products and synthesize a perfectly structured comparison is deeply satisfying. We also proved that a powerful AI tool can still respect user privacy with a local-first architecture.

What's next for BuyWise

Universal E-commerce Support: Through internal testing, our core extraction engine is actually already capable of processing most mainstream platforms. We plan to officially optimize and roll out support for sites like Walmart, BestBuy, and Shopify.

Habit-Aware Personalization: While BuyWise already provides context-specific verdicts, our next step is building long-term preference memory. The AI will learn your general shopping habits over time (e.g., prioritizing budget-friendly options, valuing durability over aesthetics) and automatically tailor its analysis to fit your unique consumer profile.

Voice-Interactive Copilot: Adding real-time voice dialogue capabilities, allowing users to converse hands-free with the AI about a product's reviews while browsing.

The "Seller Dashboard": Flipping the tool to allow merchants to use our AI pipeline to see exactly how their products stack up against competitors based on real user feedback.

Built With

Share this project:

Updates