This system enables businesses to verify the identity of agents at every stage of a transaction and allows organizations to autonomously decide which agents to transact with.Article author, source: Investing.com
GoDaddy (NYSE: GDDY), Infoblox, Google, and Cloudflare today demonstrated a system that enables AI agents to conduct transactions across independent platforms using existing internet standards.
According to the press release, the company unveiled a reference architecture called "Chain of Trust" at the GovWare 2026 conference during the Singapore International Cybersecurity Week. The architecture was jointly developed by the company and Singapore’s Infocomm Media Development Authority (IMDA), based on the principles outlined in IMDA’s Model Framework for Governance of Agentive AI.
This demonstration simulates collaborative agents operating on different platforms completing a flight ticket purchase transaction from Madrid to Singapore. The buyer agent runs on Amazon Web Services (AWS), the seller agent is deployed on the Cloudflare Workers platform, and the authorization and verification agents operate on Google's platform.
This architecture employs multiple open standards, including the Domain Name System (DNS), Agent Resources Discovery for identifying agent capabilities, Agent Name Service for authentication, OAuth for authorization, and the Agent-to-Agent (A2A) protocol for communication between agents. Each transaction is recorded in an append-only log that is verifiable by any party.
McKinsey estimates that by 2030, AI agents could facilitate global consumer spending of $3 trillion to $5 trillion.
The system enables businesses to verify the identity of agents at every stage of a transaction and allows organizations to autonomously determine which agents they wish to transact with. Users can set purchase limits and related parameters via a signed Agent Payments Protocol authorization, which merchants and payment service providers can verify.
Lee Wan Sie, Director of the IMDA AI Governance and Safety Cluster, said: “To unlock the potential of agent-based AI, we need to build a trustworthy and interoperable foundation to support responsible large-scale deployment.”
The demo also covered scenarios such as fake sellers and forged documents, with the system blocking these fraudulent activities before the transaction occurred. The simulated purchase used Stripe for payment processing.
The company has published a technical case study introducing the prototype system and invited DNS operators, domain registrars, identity service providers, payment networks, and cloud platforms to adopt this architecture. The reference architecture and case study are available at openagentcommerce.org.