- Key insight: Maximum, an AI-powered operating system startup for banks, is launching with a $30 million seed funding round to compete with incumbent core providers.
- Supporting data: Seventy-six percent of banks recently surveyed by the ABA have core systems provided by Fiserv, Jack Henry or FIS, with Fiserv alone holding core contracts with 42% of surveyed banks.
- Expert quote: "AI-based core systems hold promise but will be evaluated in terms of near-term ROI relative to infusing AI into existing systems." —Celent's Craig Focardi
As some vendors and banks are upgrading their core technology with artificial intelligence, one fintech founder is building an AI-focused bank operating system alternative.
Randy Fernando is formally launching his latest startup, Maximum, with a $30 million seed funding round led by CRV. The company is working to offer a core operating system for banks that incorporates AI directly instead of adding it to the top of existing bank cores.
The financial industry's race to adopt agentic AI includes banks looking for options to integrate the new technology with their core systems. Some major bank software vendors, such as Fiserv and FIS, have recently
Fernando is aiming to enter the bank core provider market with an offering that is built on agentic AI. His goal is to have banks use Maximum's product to internally build custom agents for unique tasks.
Other bank software providers, such as
"All financial institutions are in the process of infusing AI into their core banking systems because AI must work within the existing IT architecture to be effective, auditable and compliant," Craig Focardi, principal banking analyst at Celent, told American Banker. "AI must also generate repeatable results for all use cases. Today, AI-enhanced core systems provide immediate benefits to financial institutions, while AI-based core systems are experimental."
Fernando told American Banker in an exclusive interview that he is building a new platform instead of a layer to place on top of existing infrastructure. "We call it an operating system versus a core because we do more than what a typical core banking service would provide," he said. "You can have much more control over products and features, and even get to a place where products are dynamic or personalized for customers."
One example Fernando gave was a suspicious activity report, or SAR, agent that scans the Department of the Treasury's OFAC list every night and compares it to all the bank's transactions from that day. If a transaction is found between a customer at the bank and an entity on the OFAC list, the agent sends an alert to a compliance officer.
"It actually provides a recommendation to the compliance officer as well in terms of how to manage that particular instance," he said. "This happens in seconds versus the human team doing this over hours or days manually, [and] we have a human in the loop with the compliance officer that's acting on the findings." Other bank tech providers that offer AI agents to banks for AML compliance, including OFAC screening, include
Maximum is Fernando's third bank-focused startup. His first, an automatic retirement investing fintech called
"We've seen modern banking companies that have come to market over the last five or six years focus on chipping away at parts of the core banking stack," he said. "Most [of the] banks that we talked to or had partnered with in the past are most interested in working with institutions or organizations that have more capabilities than just a narrow scope of services."
Making the switch, however, may be easier said than done. Seventy-six percent of the 679 banks surveyed by the American Bankers Association for its most recent
Read more:
Should a bank ever be liable when a customer gets scammed? OpenAI models' Hugging Face breach is a red flag for bankers The top-performing banks with $2B to $10B of assets in 2025 Inside Upstart's journey to get a bank charter approval
Focardi said that few banks are demanding an AI-native core banking system today.
"The potential for such a system is worth exploring, just as cloud-native core banking systems were 10 to 15 years ago and have begun to help financial institutions modernize," he said. "Similarly, AI-based core systems hold promise but will be evaluated in terms of near-term ROI relative to infusing AI into existing systems."
JT Thykattil, vice president of research for Forrester Research, told American Banker that the banks Forrester works with are demanding improved operational outcomes but are deeply skeptical of buying AI at the core ledger level.
"Current core banking systems are deterministic, which banks require," he said. "Banks worry that 'AI-native' implies probabilistic logic operating too close to the balance sheet, raising severe regulatory and audit red flags. Regulators require financial institutions to explain exactly why a core system executed a transaction. AI-based reasoning engines are inherently harder to audit than current core banking systems."
Banks are looking for outcomes from any innovations in their core banking systems, Thykattil said, but the difference AI makes for them is based on where it sits in the architecture and whether it changes the operating model and produces measurable outcomes.
"A truly AI-native core banking system learns continuously from outcomes: a declined loan that should have been approved, a fraud flag that was wrong," he said. "Legacy cores have no native mechanism to close that loop. If the AI layer sits outside the system of record, its learning doesn't inherently improve the core's behavior over time."
The main areas where AI can provide operational gains for banks, according to Thykattil, are fraud and financial crime detection, credit underwriting for underserved segments such as thin-file consumers and operational automation like loan document processing.
"Banks have tried RPA [robot processing automation] and found its limits; AI is the credible next step," he said.
Paul Schaus, managing partner at CCG Catalyst, told American Banker that banks standardizing their real-time data to make it "clean and accessible" is a critical problem that needs to be solved for AI to be useful, regardless of whether it's added on top of existing infrastructure or replaces it.
"AI-native cores are still early, and being built from the ground up does not automatically make them better," he said. "A modern core with good data and open APIs will outperform an 'AI-native' pitch with no governance every time. Start with the outcome and the data, not the adjective."
Tom Seo, founder of venture capital firm Inverted Capital, told American Banker that core banking is not immune to the broader
"Value is accruing 'up the stack,'" he said. "Incumbents have historically provided utility to [bank] customers as the underlying system of record. Customers are demanding that software goes beyond existing as the system of record and instead 'do the work' as the system of action. Core banking is an especially compelling opportunity as it captures virtually all the context from which a bank takes downstream action."
Even as banks have reported less-than-glowing satisfaction with their core providers, according to the ABA core banking report, a majority (69%) still reported that they would likely stay with their current core even if they weren't fully satisfied with it.
"As we started to build Maximum, we committed to build for the long term knowing that this would not be an overnight success," Fernando said. "We realize we need to be patient as a part of this process, though, and we've luckily found investors that are committed to that long-term vision and believe that there's such a desperate need for a solution like this that they've signed up for the fact that we are in this for the next decade plus. I think that's the opportunity that is ahead of us when we get this right."
