Blend Insights 2026 Digital - Flipbook - Page 3
SPONSORED
CONTENT
Mortgage originations by quarter, Q1 2020 through Q1 2025 (000s)
Source: Federal Housing Finance Agenc y
4,000
3,500
3,000
2,500
2,000
pleted correctly. Now, SouthState is using
software that includes machine learning
1,500
to quickly review the file and kick back
anything that has a discrepancy, such as
1,000
two contradictory numbers.
500
Whipple notes that overall, the mort-
2020
Q1
Q2
Q3
2021
Q4
Q1
Q2
Q3
2022
Q4
Q1
Q2
Q3
2023
Q4
Q1
Q2
Q3
2024
Q4
Q1
Q2
Q3
2025
Q4
Q1
gage business is “very document-driven,”
so automating the review of paperwork
is “the lowest hanging fruit” in terms
of boosting efficiency. Automation can
also prevent situations where borrowers
During that period, SouthState invested
to ensure that a customer has uploaded
believe their application is complete, only
in new machine learning software to help
the necessary documents to get approval.
to discover later that something is still
it automate its loan application and re-
That can reduce mundane work tradition-
missing. Whipple gives the example of a
view processes. But the bank’s staff didn’t
ally done by bank staff members.
borrower providing what they think is the
have time to fully become accustomed to
SouthState hopes that automating some
correct document to verify income, only
this technology given the high volume of
of these processes will mean the institu-
to find out weeks later that the document
originations at the time. “We have looked
tion doesn’t need to go on a hiring spree
is incorrect. That back and forth between
at the last couple of years as a rebuilding
the next time interest rates change, and
the lender and consumer can create
time and an opportunity to lean into our
consumers are once again demanding
frustration. “Using AI to verify that the
technology,” Horan says. “We are prepar-
more mortgages. “We are looking at it as a
borrower provided the right document
ing for when, inevitably, interest rates go
way to be flexible as the market changes,”
up front, as they are uploading it, can cut
lower again.”
Horan adds. “Instead of having to hire
down on that back-and-forth cycle,” he
more people and then lay them off, we are
adds. “Banks should really think about
cyclical, being largely driven by interest
trying to fill that gap without having to
how to use technology to reduce friction.”
rates. When rates are down, consumers
change staffing levels.”
The mortgage business is famously
rush to refinance existing mortgages or
seek out a loan to buy a new home. In
response, many banks go through the pro-
Horan emphasizes that banks should
be deliberate about how they adopt AI,
A Better Customer Experience
Horan compares the mortgage origina-
rather than hopping on the AI bandwagon without fully understanding the
cess of ramping up staffing during boom
tion business to that of a factory assembly
technology or thoroughly vetting third-
times, only to lay off employees when the
line: The application moves from the loan
party vendors. He notes that there are
mortgage market slows. “Banks have to
officer to the processor to the underwriter
many fintech companies offering sound
go through the painful exercise of losing
until a decision is made. Traditionally,
technology, but banks need to ensure that
staff,” Whipple says.
much of this process has been done
the vendor is the right partner for them.
manually by employees. That process can
One way to do that, he says, is to speak
cially AI, can help limit these wild swings
lead to a bunch of “back and forth,” he
with client references before signing a
in staffing needs. AI can aid in the review
notes, if the application is missing docu-
contract. “You have to avoid falling for the
of a mortgage application, for instance,
ments or something hasn’t been com-
smoke and mirrors,” he adds.
Whipple notes that technology, espe-
3
