GivAway

Architecture

Applying to the Zurich Insurance Challenge at Hack Zurich 2023

Inspiration

Our inspiration is rooted in the desire to make a meaningful difference in people's lives during their most challenging moments. We understand that when unexpected and unfortunate events occur, the need for top-notch insurance service becomes paramount. Providing the highest level of support to our customers in their times of need is what drives our passion and commitment.

What it does

We've developed an innovative insurance assistant that communicates seamlessly through chat. In the event of an emergency, the customer can notify their assistant, which will immediately assess their well-being and, if necessary, coordinate emergency assistance. If the customer is safe, the assistant efficiently gathers all relevant details for the specific case, creating a professional report for the insurance company's investigation.

Moreover, our assistant analyzes the customer's insurance policy, providing clear guidance on whether they are fully covered, not covered, or partially covered. Our chat system supports both text and audio interactions, ensuring accessibility for all users. It even accepts accident images, capable of analyzing damage type and severity directly from visual evidence.

But that's not all. In addition to claims handling, our system prioritizes your privacy by offering text anonymization for personal documents. It intelligently replaces sensitive information with fictional data, ensuring your personal details are secure throughout the process.

How we built it

The model incorporates multiple machine-learning models as well as other technologies for each step along the pipeline:

We created a web-interface which enables the client to interact with the advanced capabilities of the assistant.
Given a basic description from the client, the assistant is using correlation models to classify the type of the case
The assistant asks for an image of the case and using segmentation models and multi-modal encoders we further analyze the case
The assistant uses Q&A models and sentiment analysis models to complete all the relevant details.
Once we have all the details in hand we produce the final report for the insurance company and also provide a suggestion to the client using GPT.
Throughout the process we use Text-to-speech and Speech-to-text models to allow voice conversation.

Accomplishments that we're proud of

We believe that our system has the potential to enhance the quality of people's lives by offering an intuitive solution that harnesses advanced technology.