Augustus CEO says banks can’t rebuild for AI and stablecoins

Augustus Bank’s CEO, Ferdinand Dabitz, says legacy clearing banks cannot truly rebuild their cores for artificial intelligence and programmable money, as his startup moves closer to launching a US national bank designed around both.

The Office of the Comptroller of the Currency (OCC) granted conditional approval for Augustus Bank N.A. on Monday under the Guiding and Establishing National Innovation for US Stablecoins (GENIUS) Act, which created a federal framework for payment stablecoins and clarified how banks and certain nonbank entities can issue and integrate dollar-pegged tokens under federal oversight.

Augustus now plans to establish a full-service national bank in Dallas, Texas, focused on fully reserved stablecoins, AI-driven compliance and automation-heavy back-office processes. Dabitz told Cointelegraph it was just “a couple of months” from full approval and launch. However, final approval remains subject to pre-opening conditions.

The company is targeting the “broken” correspondent clearing business dominated by global banks such as Citi, arguing that incumbents cannot fully re-platform systems built for humans, not machines, that still close on weekends and rely on decades-old cores.

“The short answer is replacing them,” Dabitz said when asked whether Augustus could coexist alongside traditional clearing banks.

Augustus bets stablecoins and AI can remake clearing

Augustus began life in Berlin in 2021 as Ivy, a euro-clearing fintech that built a transaction banking platform for non-US financial institutions, fintechs and crypto firms.

Augustus received conditional OCC approval this week. Source: PR Newswire

The bank already runs euro payments and instant settlement for clients, including crypto exchange Kraken. “The clearing bank bond is truly broken,” he said, arguing there's an opportunity to “rethink it as an application and deliver something pretty terrific.”

Related: JPMorgan to launch tokenized money market fund for stablecoin issuers

Central to Dabitz’s pitch is the belief that large banks can upgrade legacy infrastructure but cannot fundamentally rebuild around AI and tokenized money. “I’ve come to the conclusion it’s impossible to re-platform a bank,” he said.

Augustus plans a three-layer stablecoin model: using stablecoins as a funding rail for payments, as a treasury and liquidity tool to release what Dabitz estimates is around $3 trillion in trapped idle capital, and as the interface layer for AI agents interacting directly with money.

He said the model could enable real-time treasury optimization and allow AI systems to become “first-class customers” of the bank, handling tasks such as liquidity management and transaction monitoring on behalf of corporates.

Competition from banking giants

Dabitz’s argument comes as major banks accelerate their own AI and digital asset initiatives.

JPMorgan Chase says it invests more than $18 billion annually in technology, including AI, and Citi reported over $6.1 billion in clearing-related revenue in Q1 alone, highlighting the scale of the incumbent profit pool Augustus is targeting.

Dabitz argues his team can still move faster because it is designing AI and stablecoin workflows into its operating model from the outset rather than retrofitting existing systems.

Related: Argentine banks testing JPMorgan’s JPM Coin to speed up settlements: Report

He also described the US banking market as structurally under-innovated, noting that banking is unusually labor-heavy compared with other major industries, with people rather than assets forming a major part of operating costs.

Pushing AI deeper into banking operations

Augustus wants to compress processes such as transaction monitoring, case handling and suspicious activity reporting from “20 hours to 20 minutes” using AI, with humans supervising the systems rather than manually performing every step.

Critics question whether a young, AI-focused bank with a 25-year-old leader at its helm can safely automate compliance-heavy operations without introducing model risk, explainability problems or operational failures.

Dabitz said that only makes the challenge “more exciting” and that the company plans to work closely with regulators and banking executives to ensure “the checks and balances and the harness for the AI to operate in a safe and sound manner.”

Magazine: Bitcoin will not hit $1M by 2030, says veteran trader Peter Brandt