Vetting an Artificial Intelligence Company
May 19, 2020—We’ve all been hearing a lot about the application of AI computer vision solutions in auto claims. But when it comes down to actually getting involved, how should insurance carriers pick the best partner?
This question was raised in a in a recent webinar on "Claims Like Never Before - Agility, Leadership and Collaboration for a Seamless Claims Experience" hosted by Insurance Nexus by Reuters Events.
James Spears, head of automotive for Tractable, shares how a carrier can properly vet a third-party artificial intelligence (AI) company. Tractable develops AI for accident and disaster recovery.
Body shops and repair shops might be facing the same conundrum. Say, for instance, your shop is looking to incorporate a photo-estimating software so customers can have a touchless estimate. Companies offering this service vary, including Bodyshop Booster, Mitchell, CCC, Web-Est and more.
Repair shops could be faced with deciding which is the best messaging feature to integrate into the system so customers can easily text the shop.
How can you be sure the software your choosing is the right fit for the shop's operations?
Spears' tips to analyzing if a company is the correct fit, can be applied to any scenario in which a shop operator is looking to partner with a new technology company.
Start with a good challenge. Bring your own images to the AI provider. These images will be ones that no one has run a software on yet. Bring out the stopwatch. You want to see that once you upload those images, within two to three minutes, a very-well vetted appraisal of damages comes back. That first notice of loss decision comes back. Is it repairable? That decision is critical and then you can go forward from there.
Do your homework. Make sure that you hold this company accountable. Research the company. The company should not take two to three hours to get back to you because during that time, there is no way to know what went on in the background.
Vetting a third-party AI provider goes back to the company's goal to reduce the periods of uncertainty for the customer. The company should provide speed, accuracy and service for the customer.
Once you get over the hurdle and do the test, the next step is to make sure the AI company is calibrated with how you do business ajnd where you do business, Spears says.
"You want that company to be integrated with you," he says in the webinar. "You want it to be integrated in your app if you have an app, integrated into your first notice of loss procedures and you want those all very well tested."
Now, you begin an "engagement period." The teams come together, decide on a set of rules, run the program through and calibrate it to how you perform.
"Hold the supplier accountable and that's how you really get a great AI supplier," he says.