Sonata Bank is preparing to ramp up AI deployment, with a focus on lending, considering recent regulatory changes that could accelerate innovation at community banks.  

The Federal Reserve has rescinded SR 11-7 — a 2011 framework for how AI and machine learning is governed in banking — and replaced it with SR 26-2, making the framework proportional to the size and model-risk profile of a bank, rather than a uniform checklist, according to a letter by the Fed. 

This change “opens up a great door for community banks to start to explore and play in this space because now the expectation is scaled to the size of the institution,” Sonata Bank Chief Innovation Officer Will Rhoads told FinAi News.  

The updated policy, put in place April 17, primarily affects banks with more than $30 billion in assets, with the Fed deeming that internal AI risk management and governance practices are often more appropriate for institutions with $30 billion or less in assets.  

“No longer is a $1 billion institution being held to the same standard as a $30 billion-plus institution when it comes to model-risk management.”

— Will Rhoads, chief innovation officer, Sonata Bank

At the end of 2025, 69 of nearly 3,900 domestically chartered banks, or roughly 2%, had assets exceeding $30 billion, according to Federal Reserve data.  

SR 26-2 explicitly excludes agentic and generative AI because they are “novel and rapidly evolving,” the Fed letter reads.  

AI at Sonata 

In addition to previous AI deployment, Sonata is exploring vendor partnerships for AI in commercial lending, Rhoads said. The $282 million bank is an especially active lender in the quick-service restaurant industry, he added. 

“We’re working on using Snowflake and AI to develop a franchise concept intelligence layer, pulling in more gross data beyond financial metrics — demographic data, foot traffic data, sales data — to help us understand the overall health of the brand,” he said. “That’s a big one where we think AI can provide a lot of lift for processing these massive amounts of data and drawing connections between disparate data sets.” 

Snowflake, a cloud-native platform, is the “production backbone” of Sonata’s AI initiatives, helping the bank focus on “data integrity, how we’re managing our data and what external sources we’re piping in to map to our data,” Rhoads said. 

“I think that is a journey that a lot of banks, especially community banks, are going to have to go on over the next 12 to 24 months if they want to take advantage of the advancements in AI,” he said. 

Further, Brentwood, Tenn.-based Sonata is exploring agentic AI to support “proactive outreach” to customers, streamlining tasks that require follow-ups, Rhoads said. Business account onboarding, for example, often necessitates “chasing down customers” for key information. 

Agentic AI is effective for automating these tasks while personalizing communication, he said.  

“For community banks, a differentiator is that feeling of personalized service,” he said. “Using agentic AI to provide those personalized communications at scale … has a lot of capability and lets us kind of stretch legs on the account opening front without having to bring in a whole new vendor and go through the process of standing up some new system.” 

Register here for the FinAi Lending Summit, set for Oct. 7-8 in Las Vegas.