AmeriSave built an AI underwriting system that allowed it to scale its business 1,200%, but there’s a catch.
AmeriSave’s story offers a case study on how companies clamoring to integrate AI run into the reality that successful projects take significant time and investment – and they need to prioritize carefully.
The CIO of AmeriSave – one of the largest privately owned mortgage lenders in the US – can’t keep up with all the requests for AI apps within the company.
While the proffered ideas would tackle various cost issues, Magesh Sarma has seen firsthand that building machine learning algorithms isn’t quick’n’easy. CIOs need to be judicious in selecting projects that will drive significant business value, and use past experiences to resist the temptation to say yes to every request.
For example, Sarma’s team spent $20 to $30 million each year for several years building one complex algorithm to perform some parts of the loan-underwriting process, according to The Wall Street Journal. Ultimately, it helped AmeriSave fund $24.2 billion in loans, up from $1.86 billion just two years earlier.
That’s a big reward, but it took significant time and investment.
There’s so much AI excitement that CIOs need to be measured, lest costs and commitment exceed the final impact.
“In general, there’s so much work to be done that there are often competing priorities,” according to the CIO of insurance firm Aflac, Shelia Anderson. One of the key factors she uses to determine which AI project to tackle next is time-to-value. For more on evaluating potential tech experiments, check out advice from former Credit Suisse CIO Radhika Venkatraman.