Celent, a global technology consulting firm, interviewed executives at 106 lenders in August 2025: 24 credit unions, 43 banks and 39 consumer finance companies. Among the 106 lenders, 73 had assets under $20 billion, including 33 with assets under $1 billion.
"For retail lending institutions, it's not a question of whether or not to invest in AI, it is how much to invest, where to invest, and how fast to adopt all AI technologies to keep up with or stay ahead of the competition," according to Craig Focardi, the report's author and a Celent principal analyst.
GenAI emerged in banking in 2022 and 2023 for software coding, market research and fund performance summaries.
GenAI's advantage was that it could layer natural language summaries on top of lending data. It can serve as an interactive, conversational assistant for making decisions.
It understands plain questions such as "show me month-over-month change in credit card loan approval rates."
Nearly half of the Celent respondents (39%) said they were using artificial intelligence/machine learning in at least some part of their consumer loan originations, and another 36% were developing such a use. Common uses are risk management modeling and decisioning.
AI use dropped to 20% for GenAI for originations, but 46% were developing a use.
Only 8% were currently using Agentic AI for originations, and 28% were developing a use. Agentic AI allows the creation of autonomous, multi-step actions with minimal human input.
The lenders surveyed by Celent identified major areas where they need AI the most.
The need most commonly cited by the lenders (68%) was to overcome the inability of their origination systems to integrate with external data and decisioning capabilities.
More than half also cited their need to improve the predictive ability of underwriting and decisioning (56%), and to make more effective use of their existing customer data and alternative data sources (52%).
Focardi wrote that many loan origination systems "are older client-server systems, have weak database management capabilities (for example, difficult to add new data fields), and do not easily integrate with data the way a cloud-native LOS does."
"After the LOS integration challenge is overcome," Focardi wrote, "lenders can then solve for the second most important issue: using data and AI to improve underwriting and other loan decisioning (such as loan pricing and fraud)."
Celent's survey found only 6% of lenders already had a GenAI plan in use by August 2025.
However, 25% had a plan that they expected to implement by the end of 2025, and 16% were working on a plan that they expected to implement by the end of 2026.
Another 36% expected to complete a plan by the end of 2025, but did not specify when they expected to implement it.
Nine percent said they were discussing GenAI but "we do not have a GenAI strategy."
Finally, only 8% said they were "taking a wait-and-see approach before implementing GenAI" at their firm.
Q&A
One could say AI-driven fintechs are having a moment in the credit union industry, with new solutions popping up left and right to assist with everything from member service to fraud mitigation. One in the news lately is Vine, which recently partnered with its first credit union, the $731 million, Houston-based PrimeWay Federal Credit Union. PrimeWay is implementing Vine's AI-powered commercial lending accelerator, designed to help deliver faster decisions on commercial loans. We recently connected with Vine CEO David Eads to learn more.
CU Times: How can AI tools help solve the pain points that currently exist within the commercial lending process for credit unions?
Eads: Commercial lending is often the last remaining analog part of the credit union. Several commercial lenders in the U.S. still use a manual process using just Microsoft Excel, Word and OneDrive. Evaluating a complex commercial loan can take a full week as credit analysts gather data, analyze financials and draft and revise credit presentations.
Modern AI tools can improve this process significantly. Advances in document analysis allow lenders to accurately read and instantly extract data from large volumes of financial information. Interactive AI tools can help surface key insights and answer complex questions, even when information is buried across hundreds or thousands of pages of documentation. AI can also assist in drafting initial credit narratives, reducing the time spent on repetitive tasks.
The result is a more efficient lending process that allows credit unions to move faster, improve consistency and ultimately provide a better experience for the businesses they serve.
CU Times: PrimeWay recently became the first credit union to offer Vine's commercial lending platform. Why did Vine decide to enter the credit union space, and why is it important for credit unions to embrace AI tools in commercial lending?
Eads: More credit unions are looking to provide commercial services to their members that own or operate businesses. Commercial lending represents a meaningful opportunity to support those members while also strengthening the institution's long-term growth.
At the same time, many credit unions are building or scaling commercial lending programs with lean teams and limited resources. AI tools can help extend those teams by improving efficiency and making it easier to manage the complexity of commercial and agricultural lending. They can also support lenders who may have more experience in consumer lending as they develop expertise in commercial credit.
Ultimately, adopting these tools allows credit unions to deliver faster, more consistent lending experiences while continuing to build strong relationships with their business members.
CU Times: The use of AI in credit decisioning has raised some ethical concerns within the financial services industry, namely the potential for bias in AI-powered lending decisions. What are Vine's thoughts on this?
Eads: This is a very real concern and a real risk for financial services. AI has the potential to introduce risk if it is used to make or influence credit decisions without proper oversight.
Vine's approach is to apply AI to the operational side of lending, improving how information is gathered, analyzed and presented, rather than replacing human decision-making. We believe strongly that lenders should remain in control of all final credit decisions.
The platform is designed to provide transparency and support validation, allowing lenders to clearly understand how information is derived and ensuring they can review and confirm accuracy. This approach helps credit unions improve efficiency and consistency while maintaining fairness, accountability and regulatory alignment. At its core, our philosophy is about putting AI in the banker's hands, not bankers in the hands of AI.