Using AI to Take Action on Survey Data in Real-Time
Insurance has always been a data-driven business. However, in recent years, insurers have seen the volume, variety and velocity of unstructured data sources increase exponentially, especially in loss control.
For example, a few years ago loss control surveys mainly consisted of questionnaires and checklists. Today smart devices are allowing people to supplement surveys by attaching photos, videos, images of labels, voice memos and more. Along with the addition of smart devices, the adoption of self-surveys and video-guided surveys due to COVID-19 has allowed carriers and loss control vendor companies to capture even more information from policyholders, agents and brokers.
While these digital technologies enabled carriers to gather more survey data, they also required underwriters to manually review this information in order to take action.
In a recent study conducted by Utilant, our AI technology reduced the time it takes to review external data sources from policyholders, agents and brokers by 50%
Using AI to Take Action from Survey Data
Advanced AI and machine learning though are allowing carriers and vendors to analyze data right at the source and take immediate action. By identifying hazards and providing recommendations as data is collected, carriers and vendors can now create more value by proactively addressing issues and providing recommendations in real-time.
Below we’ve highlighted three new tools which Utilant has released that utilize AI and machine learning to help carriers and vendors to take action from survey data:
Real-Time Video Stream Analysis
Carriers and vendors using Utilant’s self-survey module or Guide Stream 360 (video-guided surveys) can utilize our machine learning technologies to automatically label, hazard score and upload images into our survey management platform. Using our state-of-the-art computer vision systems, Utilant has begun to leverage its database of over 200 million photos to train its machine learning platform to automatically label video segments in real-time. Some of the items our AI can identify and automatically label, hazard score and generate recommendations on from videos of commercial and residential properties include:
Real-Time AI Hazard Detections
After a photo has been labeled, Utilant’s Risk Alerts will analyze the photo further to identify any potential hazards in real-time. This feature improves the accuracy of surveys and acts as a “second set of eyes” to ensure critical hazards, such as missing sprinklers and railings, non-braided hoses, Federal Pacific “Stab Lok” panels, trip and fall hazards and other issues aren’t being overlooked.
- Fire Extinguishers
- Ventilation Hoods
- Hot Water Boilers
- Electrical Panels
- A/C Compressors
- Heating Systems
- Commercial Stoves
- And much more!
Real-Time Text Extraction to Identify Equipment Recalls
Using optical character recognition, Utilant’s AI can also extract text from inspection tags on fire extinguishing systems, nameplates on appliances, maintenance records, etc., from videos and photos. This information can then be cross-referenced to recall and equipment list databases to identify any recalls or potential issues in real-time.
See our video stream analysis, hazard detections and text extraction in action below!
Additional ways Utilant’s AI analysis can impact operations at carriers and vendors include:
Reduce staff turnover and inefficiencies by automating manual tasks
Improve survey data accuracy by with automatetic AI photo labeling
Boost actuarial modeling accuracy by leveraging parsed AI data
Proactively supplement survey data with recommendations, forms, and follow up questions
Verify the accuracy and content of data provided by the insured or agent