On 19 June 2026, Singapore‑based payment tech firm CardInfoLink announced that its Agent Gateway has gone live on Agenzo, the company’s own agentic commerce platform for travel and hospitality merchants. The rollout marks the first commercial deployment in Asia where AI agents operate under an independent trust and governance layer.

The Agent Gateway is part of Affinidi’s Trust Fabric suite. It assigns each AI agent a cryptographic identity, enforces policies in real time, and creates tamper‑evident audit trails. According to CardInfoLink’s release, the gateway lets agents connect to merchants through a governed, auditable channel that relies on open standards and leaves the underlying payment infrastructure untouched.

CardInfoLink has run live merchant‑acquiring infrastructure across Asia for more than 15 years. By adding the Agent Gateway, merchants can expose their services to AI agents without overhauling existing systems. Consumers can then instruct agents to book rides, flights or hotels, and every interaction is logged in a cryptographically verifiable record.

The move comes amid rising concerns about accountability in machine‑to‑machine transactions. Experian’s 2026 Future of Fraud Forecast listed the inability to distinguish legitimate agents from malicious ones as the top fraud threat for businesses that year. Gartner warned that more than 40 % of agentic AI projects may be cancelled by the end of 2027 due to escalating costs, unclear value and inadequate risk controls.

In January 2026, Singapore’s Infocomm Media Development Authority (IMDA) released the first binding governance framework for agentic AI. The framework requires bounded AI agent autonomy, human accountability and technical controls. The Agent Gateway was engineered to satisfy those requirements.

CardInfoLink’s co‑founder Hyman Zhu said the platform “provides a trusted, standardised way for an AI agent to interact with a merchant. We run real infrastructure with real merchants and real consequences, so we could not add agents without it.” Affinidi’s CEO Glenn Gore added that the gateway “is the layer that governs how an AI agent connects to the systems it acts on, independent of how the payment is settled. That layer has been missing from production. With CardInfoLink, it is now running, and every business putting agents to work will need it.”

Agenzo, developed by Evonet and CardInfoLink, focuses on enabling AI agents to complete payments safely and compliantly. It links agents, merchants, wallets, payment networks and acquirers. The platform has been used for travel booking, retail, EV charging, loyalty, QR payment and merchant checkout.

Affinidi, founded in 2020 and headquartered in Singapore, builds trust infrastructure for AI agents and the digital economy. The company contributes to the Linux Foundation, W3C and the Decentralized Identity Foundation, and its Trust Fabric suite is built on open standards to avoid vendor lock‑in.

The deployment demonstrates a production‑ready approach that the wider industry has been calling for: a layer that lets AI agents act inside real systems under governance built in from the start. For Affinidi, agentic commerce in travel and hospitality is the first commercial use of a trust layer that spans industries, from payments to healthcare and beyond.

At present, CardInfoLink and Affinidi have successfully integrated the Agent Gateway with Agenzo, enabling merchants to accept AI‑initiated transactions under a governed, auditable framework. No regulatory action or court proceedings are pending. The rollout follows IMDA’s 2026 framework and aligns with industry expectations for accountability in AI commerce.

Looking ahead, the companies plan to expand the gateway’s use to other verticals and monitor adoption among travel and hospitality merchants. No additional product launches or funding announcements were disclosed in the release.

Overall, the deployment marks a milestone in AI‑enabled commerce, providing a concrete example of how payment infrastructure and AI governance can coexist without compromising either.