Analysis
The idea connects agentic payments to deposit pricing, and funding stability. The scenario depends on agents having authority to compare accounts and initiate transfers, consumers granting usable credentials, receiving accounts being eligible, and transfers clearing without friction. Those are material operational and control assumptions. Banks can use the thesis to test , rate-sensitive outflows, customer authorization, fraud and liquidity plans; the commentary supplies no measured adoption, flow data or evidence that AI agents have caused bank outflows. Any implied banking effect is a scenario inference, not an observed event.
What remains uncertain
Apollo’s September 27 note is economist commentary and presents a hypothetical risk. Its yield comparison is author-reported; it does not document widespread agent-mediated transfers, a bank run or measured deposit loss. A later secondary report discusses the argument but does not establish that the scenario has occurred.