Deutsche Bank and Ant Digital write a paper on AI agent payments
An AI agent can suggest payments, but it cannot extend its authority, the paper tells.
On 9 November 2026, Deutsche Bank and Ant Digital Technologies wrote a white paper, Agentic Payments in Wholesale Finance. It is about AI agents, which are software that does tasks without help. The paper is about treasury, cross-border payments and trade finance. It wants bounded agentic execution, not free steps. Ant Digital Technologies made the agent platform that the paper uses as its example.
What the paper says
The paper names 3 uses for agents:
- Treasury: find cash shortfalls and move funds
- Cross-border payments: compare rates, routing and time
- Trade finance: read documents, find missing data, write payment instructions
It uses a structure of 3 layers from the International Monetary Fund. The AI work stays apart from the payment step.
The safeguards
- Limits that are clear and bounded, and that a person can stop
- Agent identity, with an owner
- Know Your Agent checks
What is not known
The paper gives no test results and no numbers on use. Its example is Anvita Flow, an agent platform that Ant Digital Technologies made. The paper thus shows a tool of one of the authors.
Insight: Banks and regulators now ask the same agent questions
On 2 November, the Hong Kong securities regulator told that it will write guidance on agentic AI. It will keep a person in charge. A person who spoke in London told that payment authentication must change for agents. This reads as a move to set the standard for agent identity and limits. The Ant platform is in the paper. Next, monitor if regulators name the same controls.
Sources
Posted
