Few industries are as constrained by their own foundations as banking. The core systems that hold deposits, move money, and maintain the ledger — concentrated in the hands of FIS, Fiserv, and Jack Henry — were architected in the 1970s around the logic of overnight batch processing. Fifty years of accreted patches have not changed that premise, and the consequences compound: real-time payments, always-on service, embedded finance, stablecoin settlement, and any serious application of AI to operational workflows all run against the grain of the architecture beneath them. This is not a failure of ambition on the part of banks. It is the absence of a credible alternative.

That is why we are excited to share our investment in Maximum, an AI-native core operating system for banks and financial institutions.

Maximum is not a sidecar, a ledger-as-a-service, or a thin modern layer sitting on top of a legacy core. It is a full core replacement: real-time ledgering and reconciliation, payments, accounts, compliance, risk, and fraud prevention, with AI built into the substrate rather than bolted on at the edges. The difference matters. Most of the venture-funded activity in this category has attacked a narrow slice — issuing, ledgers, one payment rail — leaving the bank to keep paying for the legacy core anyway. Maximum is going after the base itself, and the cost structure that comes with it. Built AI-native from the ground up, the platform allows banks to:

• Run in real time, moving from batch-based, overnight processing to continuous ledgering, reconciliation, and reporting.
• Operate autonomously, with compliance, risk, and fraud detection embedded in the core rather than bolted on as downstream checks.
• Consolidate the stack, replacing the legacy core and the sprawl of point vendors layered around it.

Just as important is how Maximum handles migration. Core conversions are the reason banks stay stuck: multi-year, spreadsheet-driven, and career-ending when they go wrong. Maximum uses agentic workflows to run the migration itself, with a shadow core running in parallel to the incumbent system before any cutover. Making migration low-risk is what converts widespread frustration into an actual switching decision.

What made our conviction easy was Randy Fernando. This is Randy’s second company; we backed his last one, Power, a credit card issuing platform acquired by Marqeta, where he got banks to fund credit card receivables against his own underwriting engine. He has done the two things this business requires most — built regulated infrastructure that banks trusted with their balance sheet, and earned the relationships that let a startup get in the room at all.

Maximum sits squarely in Anthemis’ thesis around the shift from legacy stacks to AI-native financial infrastructure. We believe the most valuable AI companies in financial services will own the system of record and the workflows on top of it, not sell assistance around the edges. A core banking system is the deepest version of that bet.

We are thrilled to partner with Randy and the Maximum team as they build the piping for the next hundred years of banking.