BlueOC is one of the founders of AICycle, first proven AI-powered solutions for car insurance in Vietnam.
Most motor vehicle insurance companies in Vietnam stay in red at 3 to 6 percent per year. The compensation rate is always high at 55%, accounting for more than 60% of insurance companies' revenue. The insurance processes are handled manually, relying primarily on the insurer's experience, making it hard to minimize operating costs and control claim transparency and insurance profiteering.
We've deeply researched the customer journey of car owners and the business processes of issuing and claiming insurance. After six months, AICycle has launched three products based on AI and Big Data, covering every single step from underwriting a new insurance policy to damages and claim assessment, including:
This module is a combination of AI and image processing technology to detect unqualified images like dark, dim, vibrating, and blurred. We use Deep Learning technology, data collection and model training to classify layers, which are then labeled by our internal tool with specific data on the Vietnamese vehicle market. The accuracy of the model based on the customer's test data reached over 98%.
We use segmentation technology to get the corresponding layer for each pixel. In addition, we collect and label characteristic defects to train models to identify types of defects. The module can identify over 100 vehicle parts and 5 different types of defects.
The Vietnamese market has many domestic and imported car manufacturers. Besides, decorating and tuning cars prevents accurately identifying a brand in other markets. Our modules are based on the characteristic shape of the car combined with the stamp to identify an accurate car brand and model.
We apply Optical Character Recognition (OCR) technology to recognize the license plate in image then transfer it to text and store. We optimized the open source of this module based on requirements and real data to produce the model with an accuracy of 99%+.
Furthermore, AICycle Data is well-designed to improve labeling team performance and supervise output. This guarantees clean data and allows us to set up labeling techniques based on specific requirements.
With ValueMe and BuyMe, the insurers can value the car by market price, verify the status of the vehicle with AI, easily search and quickly submit applications. Files are stored to compare damage when compensation is needed, preventing duplication of compensation.
ClaimMe helps to identify and evaluate damage using AI, estimate the cost of damage instantly like a real appraiser which reduces manual transaction, increases transparency in compensation cases and eliminates fraud.