news News from Cinnamon AI

  • case
  • press

Started implementing AI into NNN Life's sales support system for insurance agents "Flax Scanner for Insurance"

Started implementing AI into NNN Life's sales support system for insurance agents
“Flax Scanner for Insurance” is an AI-OCR system that also introduces the first AI to check reading image quality.  

- Significantly streamlined understanding of life insurance contracts and smooth unified management of insurance contracts -

Our company was commissioned by NN Life Insurance Co., Ltd. (President: Marius Popescu), which specializes in life insurance for small and medium-sized businesses, and we will work together to implement AI into a sales support system for insurance sales agents. . Actual operation began in November 2023.

When recruiting insurance, it is important for end-user customers to have an accurate understanding of the coverage they have already subscribed to. While it is necessary to analyze the information obtained and consider the most suitable insurance product, the format and content of insurance contract forms vary and are complex depending on the insurance company. In managing the contract, inputting contract details into the system was a burden.

In this project, we will use the AI-OCR system "Flax Scanner" that combines (1) document classification, (2) reading image quality check (a first for Cinnamon AI), (3) AI-OCR, and (4) confidence level calculation when reading (OCR) insurance contract documents. We provide By linking these with NNN Life's sales support system for insurance solicitors, life insurance contracts can be easily converted into data, helping to streamline recruitment operations.

NN Life’s integrated insurance policy management flow using Cinnamon AI’s AI-OCR system

Details of the newly introduced AI-OCR system “Flax Scanner for Insurance”

  • Form classification
    Among the customer's life insurance contract documents, we are targeting three types of documents that have similar items and can easily be misread: insurance policies, current status information, and design documents. The system creates three AI models that read each form, and outputs the reading results of the model with the highest confidence for each item for each item.
  • Reading image quality check AI
    Using image recognition technology, images taken with a camera and uploaded are checked for image quality (shake, distance, brightness, tilt, etc.) before being processed by AI-OCR, and images that are below a certain quality are immediately checked. By encouraging re-imaging, we reduce the number of requests for re-submission due to image quality after AI-OCR, thereby shortening the processing time and processing man-hours for the entire flow.
  • AI-OCR Achieves high accuracy through confidence calculation
    AI-OCR, which can accurately read documents without any pre-defined formats from non-standard documents without a unified format, adds confidence level calculation (whether or not it was read correctly) to improve the accuracy of highly difficult forms. Accuracy (item unit accuracy average 91.94%[1]) and can be read quickly (less than 3 seconds per page).
    - It is difficult to read camera images with large variations in quality (because camera shake, shadows, tilting, etc. are likely to occur)
    -Highly difficult as it targets insurance policies for corporations that have many items and are highly complex.

■Cinnamon AIResearch and Development
Cinnamon AI promotes operational sophistication to improve operational efficiency and customer satisfaction at customer contact points such as new policy underwriting and benefit/insurance claims payment operations, which are the core operations of insurance companies, and sales staff and call centers. I will continue to do so.
In addition to AI-OCR, we are conducting research and development to provide the following various AI technologies for the insurance industry.

・ Knowledge utilization/assessment support AI to support assessment operations that require advanced human judgment

・Fraud detection AI that detects fraudulent applications due to image/form falsification

・Context-based classification AI and conversation content optimization AI for conversation topics at customer contact points

・ Highly accurate sales proposal recommendation AI, etc.

■Business content: Artificial intelligence product business, artificial intelligence consulting business
Cinnamon AI provides AI consulting and AI products to realize advanced business AI solutions with the purpose of ``Together with AI, we will create a world full of creativity where everyone can envision a new future.'' We have a track record of providing services to many major domestic companies as a sherpa (leader) of change that creates new business structures for the AI era from the knowledge held by customers and leads to dramatic growth. We have established artificial intelligence laboratories in Vietnam (Hanoi and Ho Chi Minh City) and Taiwan, and have built an ecosystem to acquire and develop advanced AI human resources.We have established AI-OCR "Flax Scanner", natural language processing engine "Aurora Clipper", and voice recognition. We have accumulated technological capabilities that cover a wide range of unstructured data, such as the engine "Rossa Voice." Most recently, we have been supporting the promotion of digital transformation and AI utilization by connecting our unique AI-era strategic design "Harvest Loop" with our customers' purpose.

Please use this form to contact us regarding this matter.
https://go.cinnamon.ai/inquiry.html

[1] Item unit accuracy is the accuracy when the correct answer is only when all characters can be read without error for each reading item.