Know Your Agent

In partnership with

"Before your AI pays for your coffee, the bank should demand to know the agent."

This line is not a warning. It's a signal. The future of finance isn't just digital. It's autonomous. Financial services are entering an era where AI-powered agents will initiate transactions, negotiate on our behalf, execute trades, and manage compliance without human intervention. But with this future comes a fundamental question: if the AI acts for you, who vouches for the AI?

We've long had KYC (Know Your Customer) and AML (Anti-Money Laundering) as gatekeepers to trust in finance. Now we must introduce a new layer: Know Your Agent (KYA).

The Rise of AI Agents in Finance

As someone who's spent over a decade building fintech infrastructure, I've never seen a shift this significant. AI agents are moving beyond chat. They are becoming autonomous actors capable of making decisions and executing transactions. This isn't theory anymore. This is the next infrastructure layer for fintech.

In April 2025, Mastercard launched Agent Pay, an infrastructure built to let generative AI assistants securely complete purchases. Their innovation? The creation of agentic tokens, which are cryptographically secure credentials that authorize a specific AI agent to spend on a consumer's behalf. These tokens ensure that every transaction is traceable to a known, approved agent.

You can read more about Agent Pay from Mastercard's official press release.

Meanwhile, Visa has launched its own vision: Visa Intelligent Commerce. Their approach allows consumers to delegate Visa credentials to trusted AI agents. The AI shops, books travel, pays for subscriptions, and Visa ensures secure execution behind the scenes. The result? A payment experience driven by AI, but backed by infrastructure that consumers and merchants already trust.

These are not experimental showcases. They are pilot-tested platforms with multi-billion-dollar infrastructures, pointing to a future where your AI agent is your CFO, personal shopper, and compliance officer.

The Agent Economy

As this space matures, we are seeing the emergence of fintechs built from the ground up for agentic commerce. SemanticPay has positioned itself as the "Visa for AI agents." Their platform offers an Agentic API, enabling AI agents to access online services and make payments. It also introduces Agent ID and KYA protocols to anchor trust in every transaction.

Another emerging player is Skyfire, which launched in late 2024 as a network purpose-built for AI agent transactions. Each agent is issued a programmable digital wallet with preset limits, ensuring it cannot overspend or act beyond permission. The system features real-time monitoring and behavior-based flags. It is a microcosm of what agent governance could look like.

Skyfire's agent wallet model is an example of giving agents financial autonomy without compromising human oversight.

Why KYA Matters

In my experience working with digital identity systems and cross-border payments, attribution is everything. If we can't prove who made a decision, we can't trust the system.

KYA is the next logical layer in trust infrastructure. It involves issuing every AI agent with a verifiable digital identity, securely linking it to a human owner or institutional sponsor. Every action taken by that agent is cryptographically signed and can be logged, audited, and regulated. This allows for traceable attribution, so when your investment agent rebalances your portfolio or negotiates a lower loan rate, the bank knows precisely which agent did what, and under whose authority.

Read more about emerging standards around decentralized identity and agent credentials via the Decentralized Identity Foundation.

Your career will thank you.

Over 4 million professionals start their day with Morning Brew—because business news doesn’t have to be boring.

Each daily email breaks down the biggest stories in business, tech, and finance with clarity, wit, and relevance—so you're not just informed, you're actually interested.

Whether you’re leading meetings or just trying to keep up, Morning Brew helps you talk the talk without digging through social media or jargon-packed articles. And odds are, it’s already sitting in your coworker’s inbox—so you’ll have plenty to chat about.

It’s 100% free and takes less than 15 seconds to sign up, so try it today and see how Morning Brew is transforming business media for the better.

Check it out

Digital Identity: From Individuals to Agents

To build a trustworthy agent ecosystem, we must reimagine digital identity. For humans, digital identity frameworks like Self-Sovereign Identity (SSI) have given individuals control over their credentials. In the agent economy, we must extend these principles.

New initiatives like Master Control Program Identity (MCP-I) by Vouched offer digital passports for AI agents. Agents receive cryptographic IDs, and owners control the keys. A registry, much like a certificate authority for agents, logs behavior and issues revocation notices when an agent is compromised.

The aim? Every transaction can be attributed not just to a user, but to the exact agent instance that initiated it.

Use Cases in the Wild

We are already seeing real deployments of agentic AI that highlight the potential and necessity of KYA:

• Customer support augmentation: Commonwealth Bank of Australia uses Claude for financial services to answer queries and block fraud attempts autonomously.

• B2B Procurement: Mastercard demonstrated a textile importer agent that sources vendors, negotiates prices, and completes payment using a tokenized card.

• Debt Negotiation: Cambio, a fintech, created an AI agent that calls debt collectors to negotiate settlements. Trials show that 70 percent of users experienced credit score improvements within 60 days.

• Portfolio Rebalancing: Claude's financial services capabilities enable agents to run stress tests, rebalance portfolios, and forecast risk autonomously.

• Autonomous Trading: AI trading agents operate 24/7 in specific markets, executing micro-trades while adhering to strict guardrails.

These examples make one thing clear: we are entering an age where AI agents have meaningful financial autonomy. That means governance, attribution, and accountability must follow suit.

The Infrastructure Challenge

Building the payment rails for AI agents presents unique technical challenges that go far beyond traditional payment processing. As QED Investors notes, "the complexity of AI agents making payments is at least twice and maybe three times as complicated as making payments on the internet."

This complexity stems from several factors. First, agents need programmable spending limits that can adapt based on context, market conditions, and user preferences. Second, the velocity of agent-initiated transactions can be orders of magnitude higher than human-initiated ones, requiring new approaches to fraud detection and risk management. Third, agents operating across multiple platforms and jurisdictions need seamless interoperability while maintaining security and compliance.

The technical architecture for agent payments must also account for the dynamic nature of AI systems. Unlike static payment credentials, agent identities may evolve as they learn and adapt. This creates challenges around identity continuity, version control, and maintaining audit trails across system updates.

From Know Your Customer to Know Your Agent

KYA is not just a technical challenge. It's a philosophical shift. KYC was built for verifying humans. KYA is for a world where non-human actors operate with power and agency.

Here's a comparison of the two frameworks:

This shift is critical. Just as you wouldn't hand a stranger your credit card, you shouldn't let an unknown agent initiate a payment, open an investment, or execute a trade.

Regulation is catching up. The EU AI Act categorizes high-risk AI systems and includes clauses on explainability and auditability. Likely, we'll soon see formal requirements for agent logging, permissioning, and identity verification.

The Multi-Agent Economy

As AI agents become more sophisticated, we're moving beyond single-agent interactions toward complex multi-agent ecosystems. This evolution brings both opportunities and challenges for financial services.

In multi-agent scenarios, agents representing different parties (buyers and sellers, lenders and borrowers, insurers and claimants) will need to negotiate, verify each other's credentials, and execute complex financial transactions autonomously. This requires not just individual agent identity but also protocols for agent-to-agent authentication and trust establishment.

Consider a scenario where a consumer's purchasing agent negotiates with a merchant's sales agent to secure the best price for a product, while simultaneously coordinating with the consumer's financial agent to ensure the purchase fits within budget constraints and investment goals. Each agent must verify the others' legitimacy, negotiate terms, and execute the transaction, all while maintaining security and regulatory compliance.

This multi-agent orchestration requires new standards for inter-agent communication, shared trust frameworks, and coordinated audit trails. Early work on these challenges is emerging through initiatives like the Agent2Agent protocol, but much remains to be developed.

The Infrastructure Stack

To make KYA real, we'll need several key components:

1. Agent Identity Wallets: Unique credentials stored and signed per agent, providing cryptographic proof of identity and authorization.

2. Delegation Protocols: Sophisticated systems for delegating specific permissions to agents, including spending limits, transaction types, and temporal constraints.

3. Prompt and Action Logging: Comprehensive records of every decision point, enabling full auditability of agent behavior.

4. Revocation Registries: Real-time systems for instantly revoking compromised agents across all services and platforms.

5. Reputation Scores: Dynamic trust metrics that evolve based on agent performance, similar to credit scores but designed for AI behavior patterns.

These components are already being tested in systems like Skyfire's platform and emerging KYA frameworks being developed by companies focused on agent trust and verification.

Regulatory Implications and Global Standards

The regulatory landscape for AI agents in finance is still evolving, but several key trends are emerging. Regulators are beginning to recognize that AI agents require specialized frameworks that go beyond traditional KYC and AML requirements.

The challenge for global financial institutions is that different jurisdictions are taking varied approaches to AI agent regulation. While the EU AI Act provides a comprehensive framework, other regions are developing their own standards. This regulatory fragmentation could create compliance challenges for institutions operating internationally.

Financial regulators are particularly concerned about systemic risks that could emerge from widespread AI agent adoption. If thousands of agents make similar trading decisions based on similar data, could this create new forms of market manipulation or flash crashes? How do we ensure that AI agents comply with existing financial regulations around market abuse, insider trading, and consumer protection?

These questions are driving discussions about the need for international standards for AI agent behavior in financial markets. Organizations like the Financial Stability Board and the Bank for International Settlements are beginning to examine these issues, though concrete guidance is still emerging.

Predictions

In my view, the next 24 months will bring decisive turning points:

1. Regulated Agent Credentials: AI agents will be issued regulated credentials, similar to payment cards or bank licenses, with clear standards for identity, capability, and accountability.

2. Expanded Compliance Monitoring: Financial institutions will need to monitor not just their customers but also the agents acting on their behalf, creating new roles for compliance teams.

3. Agent-Specific Risk Controls: Banks and fintechs will develop sophisticated risk management tools specifically designed for agent behavior, including real-time spending controls and behavioral anomaly detection.

4. KYA Registry Networks: Industry-wide registries will emerge that allow financial institutions to verify agent identities and check reputation scores in real-time.

5. Programmable Agent Wallets: A new product category will emerge featuring wallets designed specifically for AI agents, with programmable limits, automated compliance checks, and comprehensive audit trails.

This evolution isn't optional. Consumers won't trust AI agents with their money unless they have confidence in the oversight mechanisms. Financial institutions won't process agent-initiated transactions without robust verification systems.

Security and Privacy Considerations

The KYA framework must balance transparency with privacy, ensuring that agent actions are auditable without exposing sensitive user data or business intelligence. This creates unique technical challenges around selective disclosure, zero-knowledge proofs, and privacy-preserving audit systems.

Agent security also extends beyond traditional cybersecurity concerns. AI agents can be compromised through prompt injection attacks, model poisoning, or adversarial inputs that cause them to behave in unintended ways. KYA systems must account for these AI-specific attack vectors and provide mechanisms for detecting and responding to compromised agents.

Conclusion

We are entering an era where your AI doesn't just advise you; it acts for you. It pays, negotiates, transfers, and trades. That demands a fundamental shift in financial systems from verifying who you are to verifying what acts on your behalf.

The shift from KYC to KYA redefines accountability and necessitates new infrastructure, standards, and trust layers. But it also unlocks a future of seamless, personalized, and proactive finance, where agents act not as mere tools, but as trusted extensions of ourselves.

The financial institutions that recognize this shift and invest in KYA infrastructure today will be the ones that lead tomorrow's autonomous economy. Those who wait risk being left behind as trust becomes the fundamental currency of the AI age.

In the future of money, it's not just who you know. It's who your agent is.