Switzerland’s State Secretariat for International Finance SIF just made a consequential claim: once AI agents start working with financial data, Open Finance becomes unavoidable.
The reasoning is straightforward. If financial institutions do not provide secure interfaces, clients may simply give AI systems access to their financial data through less controlled channels.
But the implications go much further than APIs.
If an AI agent can access data across institutions, analyse a client’s complete financial situation, formulate recommendations and eventually initiate actions,
What is the client actually buying from a bank?
- Information?
- Analysis?
- Custody?
- Allocation?
- Execution?
- Balance sheet?
- Security?
- Trust?
- Accountability?
Today, banks bundle most of these together. Open Finance and agentic AI could progressively separate them.
Some value may move down towards the hard infrastructure: custody, capital, liquidity, security and execution.
Some may move up towards something much harder to commoditise: understanding objectives and constraints, making consequential trade-offs, allocating resources and assuming responsibility for decisions.
And it goes beyond banking. It is also an organizational issue.
As information and analytical capabilities become increasingly available to machines, organizations will have to become much clearer about what clients actually trust them to provide, and what they are trusted to decide.
Strategy. Allocation. Security. Execution. Accountability.
Open Finance may be only the most visible piece. But the deeper question is what an organization is for when intelligence itself becomes infrastructure?