AI uses at banks, private AI and customer built AI apps - skills and regulations are the friction for wider adoption

The most common use case for agentic AI is customer service, according to 75% of banks surveyed by Capgemini. Nearly two-thirds use the technology for fraud detection, while 3 in 5 use it for loan processing and customer onboarding.

Also, 84% of apps/uses are built in house:

BNY developed its AI agents in-house, which one-third of banks also reported doing in the Capgemini report. Only 16% reported buying AI agents off-the-shelf.

As always with banks using new technologies, training people to use the new technology and getting regulatory sign-off are the barriers:

while AI agent adoption is poised for growth with 80% of financial services firms in the ideation or pilot stage of deployment, only 10% of companies have implemented the technology at scale, the report found.

One of the biggest hurdles executives face with AI agent deployment is a skills gap, 92% of banks reported, second only to regulatory and compliance challenges at 96%. High implementation costs are also emerging as a barrier to AI return on investments, the report said.

🔗 Banks turn to AI supervisors as agent use surges