Discovering how AI agents can become trusted commercial actors — without every merchant, wallet, payment provider, and identity system inventing its own incompatible approach.
Why this discovery exists
Agentic commerce is being shipped faster than the surrounding standards can absorb it. This page maps where the gaps are, which existing protocols already cover parts of the stack, and what a discovery effort should actually take on.
Key idea
Agentic commerce is not one protocol — it is a stack. Identity, mandates, payment rails, merchant locks, receipts, and dispute evidence each need to interoperate. The point of discovery is to name the layers, find the gaps, and avoid each vendor inventing its own incompatible answer.
§02 · Visual overview
Before the detail, three diagrams: the layered stack and its gaps, the cross-cutting concerns that apply everywhere, and how the ten discovery areas map onto both.
Diagram 1
The agentic commerce ecosystem has six functional layers. Most have at least one live standard. Several have competing standards. Post-purchase has almost none.
How to readTeal = live standard to align with. Purple = primitive owned by an adjacent body (FIDO, EMVCo, card networks). Red dashed = gap — where discovery work has the most leverage.
Diagram 2
Four concerns apply at every layer: fraud signaling, identity and trust, auditability, and PCI/data protection. Viewing the stack as a matrix shows where coverage exists and where it's missing.
How to readTeal cells have coverage. Gray cells are partial. Red dashed cells are gaps. The post-purchase row is almost entirely red — that's the single largest opportunity.
Diagram 3
The ten discovery areas divide into two groups: bullets that live on a single layer, and bullets that cross-cut the whole stack. The cross-cutting bullets are where the discovery output will be most original.
How to readGreen lines connect bullets to their single primary layer. Purple lines are cross-cutting. The five cross-cutting bullets (1, 2, 8, 9, 10) are the anchor deliverables — taxonomy, gap analysis, reference architecture, best practices, and boundary statements.
§03 · Detail
Below is each of the ten discovery areas in detail, with the existing standards to align with and the specific gaps to aim to close.
§04 · Adjacent authority layer
The discovery work on this page paints the commerce landscape. WAUTH is not a replacement for that work; it is a candidate lock-and-receipt layer for the moments where an agent wants to transact and a relying party must decide whether the proof is sufficient.
WAUTH defines the lock. HAPP defines the human return-point. Compatibility profiles let agentic commerce systems treat Verifiable Intent, AP2, ACP, UCP, network tokens, identity credentials, and future proofs as possible keys — while the merchant or doorkeeper still verifies a deterministic policy before execution.
A commerce agent can discover products and build carts, but the merchant still needs a deterministic point where insufficient proof is refused and bounded authority is verified.
When intent, amount, merchant, recurrence, or risk exceeds a mandate, HAPP-style step-up can bring the human back without making every transaction manual.
Agentic commerce needs evidence that distinguishes what the agent claimed from what the merchant, issuer, or network actually accepted and committed.
The WAUTH work has explored merchant-led sandbox pilots where Verifiable Intent is treated as upstream commerce evidence, WAUTH remains the merchant lock, and HAPP is invoked only when policy or risk requires a human to return. That makes the intersection testable without requiring the broader commerce ecosystem to adopt a single protocol stack on day one.
§04 · Reference
Acronyms used throughout this document, alphabetized.