Verdictron

Inspiration Picking the right product can be a huge hassle. Before buying anything, we often spend hours watching YouTube videos and reading blogs just to figure out if it’s actually worth it. Our goal? Cut down the time you spend researching and make the process way easier. We built a Product Review Assistant that lets you chat with it and get detailed, niche insights about products you’re considering. It also helps compare products that don’t have direct "Product A vs. Product B" comparison videos on YouTube. What it does Our project is a Framework for Product Reviews, designed to make product research effortless and fast. How we built it We combined multiple AI-powered tools to create a smooth and efficient product review system:

TogetherAI – Generates quick product summaries. Sambanova - Llama VLM – Handles fast screen analysis and YouTube transcript breakdowns. Gemini – Scrapes the web for relevant product info and provides citations. NotebookLM & Podcastfy – Used for podcast generation. HappyRobot – Enables phone calls and messaging for product info. Streamlit – Powers the UI. ActiveLoop – Stores data in a VectorDB for fast retrieval. How it works: Take a screenshot of any product, and our system extracts the product name and details. For a seamless experience, you can trigger our browser extension to screenshot a product you’re interested in. Our AI pulls relevant products from the screenshot, then Gemini fetches review links. All the gathered data is stored in a VectorDB for quick access. Once everything is processed, you can chat with our AI agent to get in-depth product insights. One of the coolest features is HappyRobot, which can actually call you and deliver product details in real-time, pulling info straight from the VectorDB. You can ask super specific questions, and our AI will find the exact answers—like if you ask about the battery life of the iPhone 16, it won’t just give you a video but also the exact timestamp where that topic is discussed. The whole process is built to be seamless, helping you get the info you need without endless searching.

Challenges we ran into

Integrating a lot of things together was a problem. 1 day was too less for adding so many things together (To be added based on challenges faced.)

Accomplishments that we're proud of

(To be added based on key achievements.)

Doing this work in a day is a very big achievent for us and is very proud of it.

What we learned

(To be added based on takeaways from the hackathon.)

Time management
project management
code management
using multiple available tools/sponsors instead of writing things from scratch ## What's next for Verdictron Make it shiny and make it OS software! (To be added based on future improvements.)