Why Agentic Commerce Requires Multi-Tokenisation

Artificial Intelligence Multi-Tokenisation

As artificial intelligence evolves, “agentic commerce” where autonomous agents buy, negotiate and transact on behalf of users, is rapidly moving from concept to early reality.

But while the idea sounds futuristic, one major hurdle stands in the way: how can these agents securely prove identity, authorise payments and manage transactions across multiple ecosystems?

Experts increasingly agree that multi-tokenisation, using several types of secure digital tokens for identity, payment, and authorisation, is essential to making agentic commerce both safe and scalable.

This article explains what agentic commerce is, why single-token systems aren’t enough, and how multi-tokenisation could underpin the next generation of autonomous digital payments.

What Is Agentic Commerce?

Agentic commerce describes a system in which AI agents autonomously complete commercial tasks on behalf of people or organisations.

Unlike chatbots or recommendation systems, these agents can:

• Search across marketplaces and suppliers
• Compare products and prices
• Negotiate deals or apply discount logic
• Complete checkouts and make payments
• Handle returns, subscriptions and post-sale support

In short, the agent acts as a digital “buyer” or “assistant”,  but autonomy creates new technical challenges, particularly in terms of identity and payments.

Find Your New Processor

Tokenisation: The Foundation of Secure Digital Commerce

Tokenisation replaces sensitive data (like card numbers or personal IDs) with unique, non-sensitive symbols called tokens.

• Payment tokenisation hides raw card data, replacing it with a “network” or “merchant” token.
• Identity tokenisation allows systems to verify a user or agent’s credentials without exposing private data.
• Delegation tokens allow users to authorise agents to act within specific limits (for example, allowing an AI to spend up to £100 per month on office supplies).

These tokens make transactions more secure, flexible, and revocable, which is why tokenisation is already embedded in card networks, mobile wallets and digital identity frameworks.

Why Multi-Tokenisation Is Essential

A single token can’t handle the full complexity of agentic commerce. That’s because an AI agent must navigate multiple domains, each with unique security, legal and operational requirements.

1. Cross-domain flexibility

An agent may interact with merchants, card networks, wallets, and APIs, all requiring different credentials. Multi-tokenisation lets each domain validate its part independently.

2. Security and control

Tokens can be issued with specific scopes and lifetimes. If a payment token is compromised, it can be revoked without disabling the agent entirely.

3. Delegation with limits

Separate tokens let users safely delegate purchasing power to agents without losing control. A single global token would be too risky.

4. Interoperability and compliance

Different regions, PSPs and identity frameworks can operate in parallel. Multi-token setups allow interoperability across fragmented standards (e.g. PSD2, GDPR, PCI DSS).

How Multi-Tokenisation Works in Practice

An AI agent acting for a business might use:

1. An identity token issued by the platform to prove who the agent represents.
2. A delegation token defining spending limits and categories.
3. A network payment token representing the business’s card credentials.
4. A merchant token is used to interact with a specific retailer.

At checkout, these tokens combine to authorise a transaction, all without exposing sensitive credentials or requiring human intervention.

If the business revokes the delegation token, the agent instantly loses permission to transact.

Technical & Regulatory Considerations

Implementing multi-tokenisation requires careful design across security, compliance and interoperability:

• Lifecycle management: Tokens must be created, refreshed and revoked automatically.
• Delegated payments standards: Open frameworks such as the Agentic Commerce Protocol define how agents handle consent, payments and delegation.
• Fraud prevention: Systems must distinguish between legitimate agents and malicious bots.
• Regulation: Liability, consent, and KYC obligations must adapt to autonomous transaction models.
• Standards alignment: Without common specifications, token fragmentation could slow adoption.

Card networks like Mastercard and Visa are already exploring “agentic token” models, while Stripe and OpenAI are developing shared agentic commerce protocols to ensure interoperability between payment ecosystems.

Challenges and Risks

Despite progress, multi-tokenisation raises new questions:

• Complexity: Managing several tokens increases technical overhead.
• Trust: Who is liable if an agent makes an unauthorised purchase?
• Security: Token leakage or misuse could expose sensitive data.
• Adoption: Merchants and PSPs must upgrade systems to support agentic standards.
• Standardisation: Competing ecosystems may delay universal compatibility.

Nonetheless, multi-tokenisation remains the most realistic path toward safe, scalable agent-driven payments.

The Future of Agentic Payments

Over the next few years, we’re likely to see:

• Hybrid token systems mix payment and identity tokens.
• Ledger-anchored trust frameworks using verifiable credentials.
• Agent-to-agent micropayments for subscriptions and digital services.
• Wider adoption of delegated payment protocols by major PSPs and merchants.

If successful, multi-tokenisation could become the invisible infrastructure behind intelligent commerce, allowing AI agents to transact securely and responsibly on our behalf.

Wrapping It Up

Agentic commerce represents a major evolution in how digital transactions are performed. But it can only function with secure, flexible ways to represent identity, intent, and payment authorisation.

Multi-tokenisation, the separation of credentials into specialised, interoperable tokens, provides exactly that foundation.

By addressing security, control, and interoperability in parallel, it paves the way for AI-driven transactions that are not only efficient but also compliant, transparent, and trustworthy.