• 

  The Insights.

• 

  The loading UI.

• 

  Found similar stuff for you.

Convie Build Story

Inspiration

Online shopping often feels like a trap. People buy things they regret because product pages are designed to persuade, not to inform. I kept seeing friends and family waste money due to fake reviews, misleading descriptions, and impulse buying. I wanted a tool that could give honest clarity before anyone hits “Buy Now”. That became the core motivation behind Convie, an AI co-pilot that tells you the truth about a product before you pay for it.

What I Learned

Working on Convie helped me grow in several areas:

• How Chrome extensions are structured (manifest, content scripts, background/service workers)
• Extracting product data directly from the webpage using DOM parsing
• Designing prompts that produce short, useful, non-generic AI outputs
• Handling async communication between different extension layers
• Thinking from a user’s perspective, not just a developer’s

How I Built It

Convie is a Chrome extension made up of:

• A content script that collects context from the product page
• A background/service worker that processes data and manages requests
• A popup interface that shows the final insights to the user
• An AI model that analyzes the product and generates a clear summary, risks, alternatives, and a buy/not-buy decision

Basic flow:

1. The user opens a product page.
2. The extension captures key product data (title, price, page text, reviews, etc.).
3. The data is sent to the AI model for analysis.
4. The AI returns a short, direct evaluation of the product.
5. The popup displays it in a simple, readable format.

My goal was to remove friction and provide clarity in seconds.

Challenges I Faced

• Websites are inconsistent, so extracting useful information reliably required defensive coding and fallbacks.
• Early AI responses were too long and vague; refining prompts to produce honest, concise advice took time.
• Getting the popup, content script, and background worker to communicate smoothly required restructuring message-passing more than once.
• Finding the right balance between “enough information” and “information overload” was a real design challenge.

Final Thoughts

Convie is still developing, but building it changed how I think about products. It pushed me to combine technical execution with actual user needs. More than anything, it taught me the value of building tools that protect people, not exploit them. The project is a first step toward making online shopping more transparent and less wasteful.

Built With

• flask
• gemini2.5pro
• googlesearchapi
• javascript
• python
• youtubeapi