Visa Launches AI Agent Developer SDK for Payments

Visa Opens the Door for AI Agent Payment Developers

Visa has officially launched its Intelligent Commerce (VIC) developer toolkit and the Digital Commerce Authentication Program (VDCAP) this month, giving payment developers and fintech engineers the infrastructure they need to build secure, autonomous AI agent transactions on the world's largest card network.

This is not a whitepaper or a pilot announcement. Over 30 partners are actively building in the VIC sandbox, more than 20 agent platforms are integrating directly, and hundreds of secure agent-initiated transactions have already been completed. For AI agent developers and payment engineers in the UK and globally, Visa's move signals that agentic commerce has crossed from experimentation into production infrastructure.

What Is Visa Intelligent Commerce?

Visa Intelligent Commerce is a suite of APIs, SDKs, and protocols that equip AI agents with trusted payment rails. The system combines four core capabilities:

• Tokenisation — agents receive context-specific payment credentials scoped to the customer's intent, not raw card numbers
• Authentication — Visa's Trusted Agent Protocol verifies agent identity and authorisation before any transaction
• Spending controls — merchants and consumers can set limits on what an agent can purchase, from which sellers, and up to what amount
• Privacy-aware personalisation — agents can access purchase preferences without exposing sensitive cardholder data

The architecture is designed so that AI agents can browse, buy, and manage orders on a consumer's behalf — securely, at scale, and within the compliance framework that card networks already enforce.

VDCAP: Authentication That Pays for Itself

The Digital Commerce Authentication Program launches in the US and Canada this month with a straightforward incentive: merchants who provide enriched data elements — Device ID, IP address, email, and billing address — qualify for a 0.05% fee reduction on transactions. Combine that with Network Tokens and the reduction rises to 0.10%.

For payment developers building checkout flows, this is a direct signal: better data quality equals lower interchange costs. The programme encourages the kind of structured, machine-readable transaction data that AI agents naturally produce — making agentic commerce not just technically feasible but economically advantageous.

Why This Matters for Backend Engineers

If you are building payment infrastructure with Rust, Go, or TypeScript, VDCAP means your transaction payloads need to be richer. The authentication data elements Visa now incentivises are:

1. Device fingerprint — hash of the agent's runtime environment 2. IP geolocation — where the request originates 3. Verified email — tied to the consumer's Visa credential 4. Billing address match — AVS data included in the authorisation

This is familiar territory for developers who have implemented 3D Secure 2.0 or Strong Customer Authentication under PSD2. The difference is that VDCAP rewards you financially for data you should already be collecting — and AI agents can populate these fields automatically, reducing friction to near zero.

How Visa's SDK Connects to the Machine Payments Protocol

Visa is not building in isolation. The company is actively supporting Stripe and Tempo's Machine Payments Protocol (MPP), extending it to support card-based payments on Visa's global network. Here is how the pieces fit together:

• MPP provides the open standard for agent-to-service payment coordination
• Visa's card-based MPP specification enables merchants and acquirers to accept card payments through MPP flows
• The VIC SDK implements the specification, handling tokenisation, authentication, and spending controls
• Stripe's Shared Payment Tokens provision agentic network tokens from Visa scoped to the customer's intent

The result is a full stack: an AI agent authenticates via Visa's Trusted Agent Protocol, receives a scoped network token through Stripe, initiates a payment via MPP, and settles on Visa's rails — all without the consumer touching a checkout page.

Rubail Birwadker, Visa's global head of growth products and strategic partnerships, described the vision: agentic commerce could become "the next level of one-click checkout, perhaps even without a button."

The Competitive Landscape: Three Protocols, One Developer Challenge

The agentic payment ecosystem now has three major protocol families, each backed by different network and platform combinations:

| Protocol | Backed By | Focus | |---|---|---| | Visa Intelligent Commerce + Trusted Agent Protocol | Visa, 100+ partners | Card-based agentic tokens | | Mastercard Agent Pay (MAP) | Mastercard, Stripe | Network-level agentic tokenisation | | Machine Payments Protocol (MPP) | Stripe, Tempo, Visa | Rail-agnostic open standard |

Additionally, OpenAI's Agentic Commerce Protocol (ACP) and Google's Universal Commerce Protocol (UCP) are emerging as discovery and orchestration layers — handling how agents find merchants and negotiate terms before the payment layer kicks in.

For fintech developers building payment systems, this means supporting multiple token types and authentication flows within the same integration. Stripe currently offers the most unified approach — its Shared Payment Tokens abstract across both Visa and Mastercard agentic tokens plus BNPL providers like Affirm and Klarna through a single API primitive.

What This Means for Payment Developers in the UK

The UK sits at a unique intersection for agentic commerce development. Open Banking APIs under PSD2 and the FCA's evolving framework already provide programmatic access to payment initiation — the same pattern AI agents need. UK-based fintech engineers are well positioned because:

• Open Banking maturity — the UK has the most developed Open Banking ecosystem globally, with established patterns for third-party payment initiation that map directly to agent-initiated flows
• PSD2 / SCA expertise — Strong Customer Authentication requirements have forced UK payment developers to build the exact kind of multi-factor, data-rich authentication flows that VDCAP now incentivises
• FCA regulatory clarity — the UK's approach to regulating digital payments and crypto assets provides a more predictable framework than many jurisdictions

The infrastructure stack for building agentic payment services looks remarkably similar to what experienced payment developers already work with: PostgreSQL for transaction state, Redis for session caching, Kubernetes for scaling agent request handlers, and Rust or Go for the performance-critical settlement and reconciliation engines.

The $1.7 Trillion Opportunity

Edgar, Dunn and Co. projects the agentic commerce total addressable market at $135 billion in 2025, growing to $1.7 trillion by 2030. Visa predicts millions of consumers will use AI agents to complete purchases by the 2026 holiday season.

The infrastructure is live. The SDKs are available. The economic incentives — via VDCAP fee reductions — are aligned. What remains is the engineering work: building the agent payment flows, implementing the token provisioning, handling the multi-protocol authentication, and scaling the backend systems to support autonomous transactions at volume.

Key Takeaways for Fintech and AI Agent Developers

• Visa's Intelligent Commerce SDK is live with 30+ sandbox partners and 20+ agent integrations — this is production infrastructure, not vapourware
• VDCAP launches in April 2026 in the US and Canada, offering 0.05–0.10% fee reductions for enriched transaction data
• The VIC + MPP + Stripe stack provides a complete pipeline from agent authentication to card settlement
• UK payment developers have a structural advantage thanks to Open Banking maturity and SCA expertise
• Multi-protocol support (VIC, MAP, MPP, ACP, UCP) is becoming essential — Stripe's Shared Payment Tokens offer the most unified abstraction today
• Rust and Go backend engineers building high-throughput payment systems are in peak demand as agentic commerce scales from pilot to production