Describe your project
PolicySense is an AI-powered platform designed to revolutionize the insurance experience by simplifying policy selection, enhancing consumer education, and automating form-filling processes.
In scope:
- AI-Powered Policy Guidance: Users receive 24/7 personalized support for any queries related to policies. Be it a doubt regarding a term or asking cancellation rules, we got you covered.
- Educational: The platform provides clear, understandable explanations of insurance terms, helping users make informed decisions.
- Chat With Your Policy: Our platform also allows you to chat with the current policy docs, so you don't have to read the docs to answer your doubt. Just chat with your doc and get your questions answered.
- Multilingual Support: To cater to the needs of the rich linguistic community in India, our website allows you to chat in any language.
- Automated Form Filling (Mock): Leveraging AI, PolicySense simplifies the insurance application process by auto-filling forms based on user data, in a friendly chatting interface, reducing errors and ensuring the user does not get stuck on any form field.
- Responsive Design: We have ensured that the website has a responsive design, meaning you can get all your doubts answered from the comfort of your phones.
Out of scope:
- Legal and Regulatory Advice: We do not offer legal advice or specific regulatory guidance regarding individual policies.
- Human Agent Support: The platform is fully AI-driven and does not involve direct interaction with human agents.
Future opportunities:
- Integrating our platform as a component that can be easily integrated into any website with just a few lines of code. In this way, sites like PolicyBazaar can ensure customers do not have to call an insurance agent to understand terms of a policy; or choose a policy for that matter.
- Scraping policy coverages from other websites to let our interface ensure the particular policy fits and suits the user. User can let AI suggest.
Challenges we ran into
Session management - storing context for the users
Prompt engineering - took a lot of hit and trial to get the prompts right to ensure a well working model
File management - ensuring files don't have to be sent repeatedly for the same context
Resource limit - working with 12RPM made it hard to test our projects since we needed to test using various prompts, especially since we do not have a monolith structure project