Simon Taylor Maps Four-Layer Protocol Stack for Secure Personal AI Agents in Finance

·2026.10.24 04:15

Key point

The map identifies specific standards like Visa TAP, Proof x401, and MCP that allow banks to verify agent identity and match actions to user-approved limits.

1 / 7

Details

  • Banks currently block agents using shared credentials because they cannot distinguish the agent from the user or verify specific authorization, creating a fraud and liability gap.
  • The proposed solution is a four-layer protocol stack: Entry (supported sign-in), Identity (proof of person), Tools (precise operations), and Intent (matching action to approval).
  • Entry layer relies on Personal Agent Protocol (PAP/Poppy) for discovery, Visa Trusted Agent Protocol (TAP) for agent introduction, and Web Bot Auth for operator verification.
  • Identity layer uses Proof x401 to request evidence, OpenID4VP for credential delivery, and FIDO standards for passkeys to authenticate without sharing passwords.
  • Tools layer employs MCP and WebMCP to define precise operations (e.g., freezing a card) rather than relying on browser clicks, while A2A handles task handoff without authorization.
  • Intent layer ensures the actual action matches the user's specific terms (e.g., hotel booking limits) using Mastercard Verifiable Intent, Google AP2, and Visa Intelligent Commerce.
  • JPMorgan is blocking shared-credential services while developing correct access methods, and Walmart has announced an MCP-based Agent Gateway with spending limits.
  • Consumer control requires businesses to grant access and verify signed actions, with audit trails linking permissions to agent makers and humans.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.