Cross-cutting
Personas
Nineteen seats at the table, from the delegating consumer to the agent itself. Open a persona to read every protocol from that seat, or scan the heatmap to see who each protocol matters to.
The personas
- Delegating consumerWants an agent to buy for them without losing control, refunds or privacy.
- Agent developerBuilds shopping or API-consuming agents; wants clear SDKs, errors and test modes.
- AI surface product leadRuns a chat or search surface (ChatGPT, Gemini, Copilot); wants conversion, trust and clear liability.
- Small merchant on a platformWants agent sales with near-zero integration work and no new fraud.
- Enterprise commerce architectWants control, merchant-of-record status, customer data and predictable integration cost.
- API, data or compute sellerWants per-request or metered payment with no accounts and low fees.
- PSP or acquirer product leadTokens, routing, risk, fees, and support for many protocols at once.
- Card issuer risk and product leadAuthorization decisions, fraud, disputes, Regulation E exposure.
- Card network product ownerToken standards, rules, ecosystem programs.
- Wallet or credential providerHolds credentials, captures consent, issues tokens (Google Pay, PayPal, Coinbase).
- Payment facilitator or relay operatorVerifies and settles, pays gas, screens sanctions.
- Validator, auditor or attestorRe-execution, zkML or TEE attestation, reputation services.
- CDN or bot-management engineerVerifies signatures at the edge and separates good agents from bad bots.
- Fraud and risk analystSignals, scores, chargebacks, agent abuse patterns.
- Compliance officer or regulatorKYC, KYB, KYA, AML, consumer protection, AI rules.
- Security researcher or red teamerBreaks protocols and implementations, writes advisories.
- Agent controls platform builderThe merces.ai team builds the controls platform for agentic transactions: a Risk Check, a registry of real agents, monitoring and policy enforcement across protocols.
- The agent as an economic actorDiscovers, pays, proves and gets rated; the machine persona.
- Academic researcherMeasurement, economics, formal analysis.
Persona by protocol relevance
Each cell is computed from the protocol page's persona section: High when the persona has its own card there,Medium when it owns, implements or verifies a step in the responsibility table, Low when it only relies on or observes a step. Rows are sorted by total relevance.
- High: A main persona on the protocol page
- Medium: Owns, implements or verifies a step
- Low: Relies on or observes a step
- None: No role found in the protocol page
- No data yet: The protocol persona file is not written yet