Taiwan's OEN Technology Builds 'Last Mile' Payment Infrastructure for AI Commerce — Trust Economy Holds the Key to the Agent Era

A Taiwan Startup Eyes the "Last Mile" of Payments

OEN on the Last Mile of AI Commerce - Taiwan Business TOPICS

AI agents may soon initiate transactions on behalf of users, but the final payment step still requires trusted infrastructure. OEN Technology is building systems to power that last mile of AI commerce.

The era of AI agents searching for products, comparing options, and recommending optimal choices has already arrived. However, the next step — actually moving money — remains a significant challenge. Taiwan-based startup OEN Technology is building payment infrastructure focused precisely on this "last mile."

OEN Technology was founded in 2020 by Hsiao Hsin-cheng, a former software engineer who was active in New York's startup ecosystem. After returning to Taiwan in 2018, he ran for city council. Through that campaign experience, he discovered a structural gap: trust-based organizations such as NPOs, religious groups, and political organizations lacked adequate digital payment tools.

Industry Context

Payment infrastructure development in agentic commerce has been accelerating rapidly since the latter half of 2025.

In September 2025, Google Cloud announced the "Agent Payments Protocol (AP2)". AP2 is an open protocol for AI agents to securely initiate payments. Over 60 companies including Adyen, American Express, Mastercard, and PayPal have joined the initiative, functioning as an extension of the Agent2Agent (A2A) protocol and Model Context Protocol (MCP). At AP2's core is the "Mandate" — a cryptographically signed, tamper-proof digital contract that records user intent as explicitly authorized in a traceable manner.

In October of the same year, Visa announced the "Trusted Agent Protocol (TAP)". TAP is built on the HTTP Message Signature standard and provides a framework for establishing trust between merchants and AI agents. Visa predicts that millions of consumers will complete purchases through AI agents by the 2026 holiday season.

As global payment giants race to support AI agents, OEN Technology is entering this market from a unique angle.

The Unique Position of the "Oen Economy"

What sets OEN Technology apart from other payment providers is its specialization in non-commercial transaction domains known as the "oen economy." The word "oen" derives from the Japanese word meaning "support/cheer," referring to trust-based transactions without commercial consideration — such as donations, political contributions, and religious offerings.

In an interview with Taiwan Business TOPICS, Hsiao stated, "Most software is built for commercial transactions. But when you're asking people to support a cause, a candidate, or a community, the nature of the transaction is fundamentally different."

After five years of growth, OEN's platform has evolved into an integrated system combining payment processing, supporter management, compliance tools, and outreach capabilities. It is currently used by political organizations, NPOs, religious institutions, university alumni donation campaigns, music festivals, and independent artists. It has been adopted for multiple years at large-scale events such as Kaohsiung's Megaport Festival.

The company also prioritizes investment in security and governance, having obtained PCI DSS Level 1, ISO 27001, and ISO 27701 certifications.

The "Trust" Required of Payments in the AI Agent Era

Even as AI agents become capable of determining "what to buy," "when to buy," and "from whom to buy," they cannot legally move money in regulated markets like Taiwan. Hsiao points out that "the final step requires a licensed payment operator working under local regulations."

This challenge is even more acute for non-commercial transactions. "In the case of donations, there is no market transaction to justify the payment. From a regulatory perspective, compliance becomes even more critical," Hsiao explains.

The "standardized expression and verification of user intent" provided by Google's AP2 is particularly important for transactions requiring high transparency. In areas such as political contributions and religious offerings — where misunderstandings can lead to serious consequences — mechanisms proving that an AI agent's actions were "explicitly authorized" are essential.

Hsiao currently serves as chair of the Innovation and Security Committee of the Taiwan Third-Party Payment Association, working to coordinate between the private sector, regulators, and global technology companies to make protocols like AP2 viable.

Impact and Applications for E-Commerce Operators

AI agent-driven payments are becoming a trend that e-commerce operators cannot ignore. According to PYMNTS.com, AI agent-led commerce is transitioning from the experimental stage to full-scale adoption. The agentic economy is projected to reach $3-5 trillion globally by 2030.

Here are the key points e-commerce operators should consider now.

Reassessing payment infrastructure: In a world where AI agents initiate transactions, the protocols supported by payment providers (AP2, TAP, etc.) become critical selection criteria. Operators need to verify whether their payment partners support these standards.

Balancing bot prevention: According to Visa's research, AI-driven traffic to retail sites has surged over 4,700%. Systems that distinguish legitimate AI agent transactions from malicious bots need to be established.

Looking beyond commercial transactions: OEN's case demonstrates that demand for AI agent payments is growing in donation and fundraising domains beyond e-commerce. This is a relevant development for businesses with crowdfunding or subscription-based support models.

Conclusion

Hsiao states, "Speed matters, but safety matters more. In a trust-based economy, moving carefully is not weakness — it's responsibility."

In an era where AI agents act autonomously, licensed payment operators are not "replaceable intermediaries" but rather "essential infrastructure" that ensures autonomy does not compromise accountability. Taiwan possesses a robust regulatory framework, a mature payment ecosystem, and use cases demanding high trust and transparency — positioning it as a potential testbed for the next chapter of AI commerce.

Who will own the "last mile" of AI commerce? The answer to this question will define the shape of commerce in the agent era.