Introduction to Automobile Insurance Claim Processing with Amazon Bedrock recipe

The recipe facilitates the entry of necessary details into the incident claim form, allowing you to upload up to five images with a total size limit of 5 MB. It is recommended to upload photos of the damaged vehical from all sides for the Large Language Model (LLM) to work better.

Upon submission, the process generates a claim ID and initiates vehicle verification by validating the provided information against the dataset. After successful validation, the process retrieves a sample price list and part details corresponding to the selected vehicle.

The process then checks the image format before proceeding with damage recognition using the specified LLM. This model assesses the extent of damage, categorizing it as either simple or complex. LLM recognizes damage only from the list of supported parts. If the damaged parts are not in the list, LLM returns the damage level as complex.

For simple damage levels, the LLM provides a list of damaged parts, and the process calculates an approximate payout.

For complex damage levels, the application undergoes further review to determine the list of damaged parts based on the uploaded images. An email containing the images and claim information is sent to the reviewer.

After the reviewer's assessment, the process calculates the approximate payout and sends an email to the specified recipient with the payout details.