CoverClever chatbot for answering detailed questions about insurance

CoverClever: Making Insurance Information More Accessible

Inspiration

Navigating health insurance can be frustrating, with complex policies, scattered information, and hard-to-decipher jargon. We realized that even savvy individuals struggle to get clear, accurate answers about their coverage. We at CoverClever will fill this gap and have a large language model powere by GPT answering any and every question you have bout your insurance policy

What it does

CoverClever has scraped data from BlueCross BlueShield and feeds it into an LLM-powered chatbot. Users can ask detailed, context-aware questions via a browser extension, receiving clear and relevant answers instantly at their fingertips

How we built it

Data Scraping: We extracted structured insurance data from BlueCross BlueShield using web scraping techniques in a JSON format
LLM Integration: The scraped data is processed and fed into a large language model, enhancing its contextual understanding.
Browser Extension: We built an intuitive browser extension to let users query the chatbot seamlessly while browsing insurance-related content.

Challenges we ran into

Data Structuring: Extracting and formatting insurance data into a structured, usable format was tricky.
LLM Fine-Tuning: Ensuring the AI provided accurate, context-aware responses without hallucinations required careful prompt engineering.
Browser Extension Integration: Embedding the chatbot into a lightweight and responsive browser extension posed UI/UX challenges.

Accomplishments that we're proud of

Successfully scraping and structuring complex insurance data.
Integrating a chatbot that delivers precise, context-driven answers.
Creating a functional and user-friendly browser extension.

What we learned

The importance of structuring scraped data for AI-powered applications into a cerrtain format.
Best practices for prompting our LLM API for the best possible answerss
Optimizing browser extensions for speed and usability.

What's next for CoverClever

Expanding beyond BlueCross BlueShield to other insurance providers with seamless and real time scraping
Have user enter preferences seperately and asking the LLM to score the provided insurance policy based on the preferences set by the user
Expanding and covering other types of insurance