MetaComp Pte. Ltd., a Singapore-licensed major payment institution processing over $10 billion in transactions, launched the StableX Know Your Agent (KYA) Framework at Money20/20 Asia in Bangkok on April 21. The framework establishes governance rules for AI agents operating in regulated financial services across payments, compliance, and wealth management, covering how agents are identified, authorized, monitored, and held accountable, according to PR Newswire.
The Problem KYA Addresses
AI agents are already initiating payments, making compliance decisions, and managing portfolios in financial services. The governance gap: there is no agreed standard for agent identity, permissions, or accountability when agents act outside their mandate.
“When a human leaves an organisation, their access is revoked. When an AI agent completes a transaction, its identity and permissions do not automatically expire,” MetaComp Co-President Tin Pei Ling said in the announcement. “It can persist in a system long after its mandate has lapsed, with no verified identity anchor, no accountability chain, and no mechanism to intervene.”
The scale of the problem: fewer than one in three organizations have adequate governance and controls to oversee AI agents, according to McKinsey’s 2026 State of AI Trust survey, as cited in the announcement. In Singapore specifically, PwC’s Global AI Performance Study 2026 found that only 47% of businesses have a documented responsible AI framework, compared to 63% among global AI leaders, despite Singapore businesses reporting higher AI adoption rates.
Four Governance Pillars
KYA organizes agent governance into four pillars, per the announcement:
Agent Identity and Registration. Every agent is anchored to a verified identity linked to a real-world individual or institution through a tamper-resistant registry.
Authority and Permission Control. Agents operate within strictly defined permissions governing what they can access, decide, and execute, with built-in safeguards requiring human escalation when actions exceed approved thresholds.
Behaviour Monitoring and Risk Intelligence. Continuous real-time monitoring assesses not just what actions agents take, but how they execute and whether outcomes align with intent. Risk profiles update dynamically.
Ecosystem and Interaction Governance. Agent-to-agent interactions are governed by extending the FATF Travel Rule, requiring exchange of verified identity and transaction information across agent-initiated activities within a unified architecture.
The Regulatory Foundation
MetaComp developed KYA drawing on Singapore’s IMDA Model AI Governance Framework for Agentic AI, published in January 2026. The company says it went back to IMDA directly for feedback and is in active engagement with other regulators, per the announcement. Singapore’s Budget 2026 established a National AI Council chaired by Prime Minister Lawrence Wong, designating finance as one of four national AI mission sectors with committed regulatory sandboxes.
The Skill Ecosystem
Alongside KYA, MetaComp expanded its AgentX agentic financial services Skill ecosystem. The first Skill, VisionX Know Your Transaction (KYT), packages compliance screening into a single agent-callable layer combining more than four blockchain analytics vendors in parallel. The skills are available through Claude, Claude Code, OpenClaw, and other compatible AI platforms via Model Context Protocol (MCP). Additional skills for cross-border payments, treasury, and wealth management are planned for late Q2 2026.
Why Singapore Keeps Leading
Singapore’s regulatory approach to agent governance is developing faster than other jurisdictions. The IMDA framework, the National AI Council, and now a licensed financial institution publishing an open governance standard within the same quarter suggest a deliberate strategy: establish norms before agent-driven financial services scale, not after. For fintech teams operating across jurisdictions, KYA’s FATF Travel Rule extension to agent-to-agent transactions signals where compliance expectations are heading.