- Product truth: If the catalog is inconsistent, an agent’s selections will look arbitrary (“the fallacious shirt,” “the fallacious size,” “the fallacious material”), and trust collapses quickly.
- Payee truth: Agentic commerce expands beyond cards to account-to-account and open-banking-connected experiences, broadening the universe of payees and the necessity to recognize them accurately in real time.
- Identity truth: People operate in multiple contexts (work versus personal). Devices shift. A system that can’t distinguish amongst these contexts will either block legitimate activity or approve dangerous activity, each of which damage adoption.
Because of this unified enterprise data and entity resolution move from nice to must operationally required. The more autonomy you wish, the more it’s essential to put money into modern data foundations that ensure it’s secure.
Context intelligence: The missing layer
When leaders discuss agentic AI, they often deal with model capability: planning, tool use, and reasoning. Those are vital, but they should not sufficient.
Agentic commerce also requires a layer that gives authoritative context at runtime. Consider it as a real-time system of context that may answer immediately and consistently:
• Is that this the fitting person?
• Is that this the fitting agent, acting inside the fitting permissions?
• Is that this the fitting merchant or payee?
• What constraints apply straight away (budget, policy, risk, loyalty rules, preferred suppliers)?
Two design principles matter.
First, entity truth should be deterministic enough for automation. Large language models are probabilistic by nature. That is useful for creating options for writing and drawing. It’s dangerous for deciding where money goes, especially in B2B and finance workflows, where “probably correct” just isn’t acceptable.
Second, context must travel on the speed of interaction and remain portable across the complete connected network value chain. Mastercard’s experience optimizing payment flows is instructive: the more services you layer onto a transaction, the more you risk slowing it down. The pattern that scales pre-resolves, curates, and packages the signal in order that execution is lightweight.
This can also be where tokenization is heading. Initiatives like Mastercard’s Agent Pay and Verifiable Intent signal a future wherein consumer credentials, agent identities, permissions, and provable user intent are encoded as cryptographically secure artifacts — enabling merchants, issuers and platforms to deterministically confirm authorization and execution at machine speed.
