The next Banking Channel Is Your Customer's Workflow
OpenAI's Codex announcement is interesting for a reason that has less to do with coding and more to do with who is doing the building.
The headline isn't that AI can write code. It's that OpenAI is increasingly designing these tools for analysts, marketers, operators, researchers, and bankers. The people closest to the business problem are starting to build the solutions themselves.
To be clear, we are not yet at the point where a non-technical user can write a prompt and reliably ship production-ready software. Anyone actively using Codex, Claude Code, or Cursor knows there is still a real gap between a useful prototype and an enterprise-grade application.
But compared to six months ago, and certainly compared to a year ago, the pace of progress is hard to overstate. The direction is clear.
Business users are beginning to build their own workflows, automations, and AI agents. And increasingly, those workflows need to touch financial systems.
That is the part I think financial institutions should be paying attention to.
The AI ecosystem won't be defined only by the applications banks build for themselves. It will be defined by the thousands of workflows their customers build on their own. Finance teams will stand up treasury agents. Operations teams will automate payables and receivables. Businesses will build approval chains, reconciliation, and payment flows shaped around how they actually run.
The question won't be whether those workflows exist. It will be whether the bank can safely and securely participate in them.
The institutions that offer a trusted connectivity layer into customer-built workflows will deepen relationships, open new distribution channels, and stay embedded in how financial work gets done.
We are already seeing the signal. Mercury, a banking fintech, has grown to more than 300,000 customers, predominantly made up of businesses and startups, since launching in 2019, with roughly $650M in annualized revenue and four straight years of profitability. What stands out now is where the growth is coming from. New customer applications are up 2.5x year over year, a fast-rising share of those new customers are AI-native companies, and their founder has put it plainly: AI is collapsing the gap between having an idea and having a company. The customers Mercury wins are the ones building, and they've made their choice based on where their tools can actually plug in.
That is the shift. The future distribution channel isn't the banking portal. It's the workflow.
Banks that get ready to participate now will sit at the center of how the next decade of business actually runs.
The programmable banking era isn't coming. It's here.