Artificial intelligence is already reshaping collision claims in ways many people inside the industry do not yet fully recognize.
That was the message from Mark Fincher and claims executive Bill Brower during a recent webinar hosted by Collision Industry Electronic Commerce Association, where both speakers argued the technology is spreading far beyond photo-based estimating.
“Everybody thinks about AI for taking photographs and creating estimates through AI,” Brower said. “But there’s other ways that the industry is using AI that the audience may not think about very often.”
Those uses now include first notice of loss intake, claim triage, total loss decisions and payment processing, with artificial intelligence quietly embedded across the claims workflow.
Brower said insurers are increasingly using AI systems to handle intake calls, collect loss information and guide claimants through the first steps of the process.
“I’ve been very impressed with how this capability has improved a lot,” he said.
He described ordering takeout from a restaurant and realizing the call was handled entirely by an automated system.
“I was very… concerned this was going to be terrible,” Brower said. “But it actually went smooth. To be honest with you, it went smoother than if I talked to a person.”
In claims operations, similar tools can ask policyholders questions, collect loss details and route the claim to the appropriate channel.
Artificial intelligence is also being used to evaluate early damage signals from photographs, allowing insurers to triage claims quickly and identify potential total losses before the vehicle reaches a repair facility.
“You can just take a couple of pictures and send it,” Brower said. “The AI will determine, is this a low severity accident that needs to go to a shop, or is it likely a total.”
Other applications include payment processing. AI can pull information directly from claim systems and populate payment workflows, reducing manual processing previously required for cheque issuance or bank transfers.
While estimating has become the most visible use case, Fincher said it is only one piece of a much broader transformation.
“If you just take that average claims adjuster… they can have up to 100 claims open at any given time,” he said, noting that AI tools can help review documents, summarize files and highlight the most urgent items.
He said the technology performs particularly well at repetitive, high-volume tasks.
“It can review photos really quick,” Fincher said. “It can identify damage on those photos, help classify claims and route claims to the appropriate place.”
But the speakers emphasized that artificial intelligence still requires human oversight, particularly in collision repair where hidden damage often cannot be identified from images alone.
“A photo doesn’t always tell the full story,” Fincher said. “There could be a lot of non-visible damage that the photo can’t see.”
Human judgment remains essential for validating estimates, confirming repair operations and ensuring the final repair reflects the actual condition of the vehicle.
Both speakers rejected the idea that AI will eliminate estimating or adjusting jobs, instead describing it as a tool that removes repetitive work and allows staff to focus on higher-value tasks.
Brower referred to the approach as “human-guided AI.”
“What we did is this process actually made the employee’s job easier,” he said, describing a bodyshop example where artificial intelligence helped generate estimates within minutes while the customer was still at the facility.
Previously, estimates might be written hours or days later after staff reviewed photos.
With AI assistance, the estimate can be produced within 15 to 20 minutes, allowing the repair to be scheduled before the customer leaves.
“That night homework went away,” Brower said.
The result in that example was a 17 percent increase in revenue and a 25 percent reduction in cycle time.
Fincher said the rapid progress in artificial intelligence over the past decade is tied to broader digital transformation across the collision industry.
“In 2010 our research showed there was less than 4,000 shops using a shop management system,” he said. “Today there’s over 15,000 collision repair facilities using a shop management system.”
That digitization created the data foundation required for machine learning.
Fincher said AI models have now been trained on more than $1 trillion in claims data and hundreds of millions of images, allowing systems to detect patterns and make recommendations based on historical repair outcomes.
Advances in computing power and cloud infrastructure have also made it possible to run complex models at scale.
He added that interoperability between industry platforms will be critical as artificial intelligence expands across the claims ecosystem.
“AI only creates real value when it can work across the ecosystem, not just in a silo,” Fincher said.
Despite the momentum, Brower warned insurers and repairers against relying on automation without providing a human fallback.
He described calling an insurance company about a claim and being routed correctly by an automated system that recognized his phone number and the open claim file.
The problem, he said, occurred when the system transferred him to an unavailable adjuster.
“I couldn’t get out of the loop,” Brower said. “I couldn’t get out and talk to a person.”
For companies implementing AI, he said, the lesson is to design systems that keep people in the loop.
“You’ve got to have the option for the human,” Brower said.
Used correctly, both speakers argued, artificial intelligence should free collision professionals from repetitive administrative work and allow them to focus on customer communication, repair quality and safety.
“Taking the low-value work out and escalating the higher-value decisions to the human,” Fincher said, creates more time for professionals to work directly with customers and technicians.
Instead of sitting behind a desk writing estimates, he said, estimators can spend more time with vehicle owners assessing damage and explaining the repair process.
“That’s such a better experience,” Fincher said. “We can show empathy during that process.”
