Examining the Potential of AI for Estimating
The collision repair industry is undergoing a transformation, and new technology is being introduced at a rapid pace. And, often, the introduction of new tech leads to skepticism—is it simply a shiny new object, or will it alter the way people work?
Jason Verlen, senior vice president of product management at CCC Information Services Inc., is very aware of that line of thinking as it pertains to CCC’s Smart Estimate tool, which was launched in late 2018. The software uses estimating logic and artificial intelligence (AI) photo analytics capabilities to write an initial estimate on a vehicle automatically. The adjuster can then look over the estimate to sign off on it or make tweaks, as necessary.
As of now, the product is only available on the insurance side, but CCC has long-term plans to extend the service to body shops, Verlen says, with the hope of producing fewer supplements, cutting down estimating time, and potentially eliminating totaled vehicles from arriving at shops.
So, shiny new object, or work-altering tool? Steve Trapp, strategic accounts manager in North America for Axalta Coatings Systems, says he is still weighing the true usefulness of AI in the collision repair space. While no one can predict the future, he says, if AI helps to identify just 80–95 percent of the damage, it would save significant time in estimating and potentially prove very valuable to shops.
But it all comes down to correct use, and shops not using technology as a means to bypass attention to detail. If estimators simply accept whatever the tool tells them, Strapp suggests, then it could be detrimental to the business.
“I can see both sides,” Trapp says.
How It Works
Prior to the introduction of the Smart Estimate tool, CCC already offered AI technology in the claims process. In its AI-powered workflow, a customer can upload a single photo of the vehicle’s damage. CCC’s system can then analyze the photo using AI to determine if the vehicle was potentially a total loss or repairable. If the car was deemed repairable, the customer would then be asked to take additional photographs of specific areas of the vehicle, and from specific angles, at the request of the CCC system. Those photographs get sent back to the insurance company, and the adjuster could then use those to build an initial estimate.
Time Savings at Each Step
Mike Cox, owner of Ed’s Collision & Glass in Maple Grove, Minn., says when he contemplates if such an estimating tool would be useful for his shop, he is more interested in knowing how affect how he trains his estimators and how accurately each estimator does the job.
If the tool can walk through each piece of the car that is (and isn’t) damaged, then Cox says it could be helpful. Otherwise, such a tool could hinder the time he uses to teach his estimators how to properly look for damage on the vehicle with their eyes and understand how damage on the exterior can affect the interior of the vehicle. The tool would potentially make it easier for estimators to slack in looking for damage.
If a person is driving around all day, getting out of the car and doing an estimate by hand, on
average the person will do about four per day. Conversely, if a virtual estimate is done, then adjusters tend to write about 14 per day, according to Verlen.
The tool was designed to be more configurable for user, Verlen says. For example, the adjusters now can set the tool to include high probability items like parts so that the person does not have to find these items and can focus on the remainder of the estimate. The person can also set the tool so that the adjuster can select among specific parts like OEM, aftermarket or recycled.
“AI is new in this domain and there likely healthy skepticism around it,” Verlen says. “But, as professionals interact with it, this builds trust.”