The open-source payment standard Lobster.cash, created by Crossmint in partnership with Visa, Solana, Circle, and Stytch, announced on April 16 that it has integrated with Mastercard, giving AI agents the ability to autonomously make purchases over Mastercard's global card network. The integration connects AI agent software to virtual Mastercard credentials with programmable spending limits, allowing agents to pay for cloud services, API access, and SaaS subscriptions without requiring human approval for individual transactions. The integration marks a concrete move toward infrastructure that treats AI agents as independent financial actors rather than tools supervised at every step.

Lobster.cash issues AI agents virtual cards on both Visa and Mastercard networks, alongside USDC (a dollar-pegged stablecoin) on the Solana blockchain for blockchain-based settlements. The design keeps raw card credentials and private cryptographic keys entirely outside the agent's operational context. Instead, agents request payment actions through a structured interface, a system intended to prevent a class of attacks called prompt injection, where malicious instructions embedded in content can manipulate an agent into leaking or misusing sensitive data. A research paper published on April 13 by teams from UC Santa Barbara, UC San Diego, Fuzzland, and World Liberty Financial identified 26 compromised AI routing systems that were secretly injecting malicious tool calls to intercept payment data, and documented one case in which a $500,000 Ethereum wallet was drained through a manipulated router. Lobster.cash's credential-isolation approach directly targets that vulnerability.

The Mastercard side of the integration runs through the company's Agent Pay program, which operates on two mechanisms. Agentic Tokens are per-agent payment credentials derived from a single underlying card account; each token carries scoped rules covering spending categories, dollar limits, and time windows. Verifiable Intent is a separate open standard that creates a signed, tamper-resistant record of what a user authorized before an agent acts. Mastercard describes Verifiable Intent as establishing "a shared source of truth across the ecosystem" when agents transact on behalf of humans. Users can revoke or restrict any individual agent token without affecting the underlying card.

Lobster.cash emerged from the OpenClaw project, a 24/7 autonomous agent framework originally developed by Peter Steinberger under the name ClaudeBot. More than one million OpenClaw agents have been deployed as of early 2026, creating concrete demand for dedicated payment infrastructure. The payment standard was initially launched in February 2026 under the name ClawPay. Crossmint, the company behind the infrastructure layer, raised $23.6 million from Ribbit Capital, Franklin Templeton, and Lightspeed Faction in 2025 and supports both Visa Intelligent Commerce and Mastercard Agent Pay rails through a single developer integration. Agents route payments to whichever rail is faster, cheaper, or more widely accepted for a given transaction, a design choice that lets developers build once and reach both networks without maintaining separate integrations. A community-created token called CLAW trades on Solana with a market cap of approximately $103,000 and roughly $16,000 in 24-hour volume as of April 15, 2026, but it carries no official affiliation with the OpenClaw project or Lobster.cash and should be treated as a speculative memecoin rather than a project investment.

Regional reach: India is furthest along, Africa faces structural gaps

Mastercard's agentic commerce rollout has moved quickly in South Asia. In February 2026, the company completed India's first authenticated agentic commerce transaction in New Delhi, working with Axis Bank, RBL Bank, Cashfree Payments, Juspay, PayU, and Razorpay on the payments side, and Swiggy, Instamart, Vi (a telecom operator), and Tira (a retail merchant) as merchants. India's existing digital payments infrastructure, built on the Unified Payments Interface and a large developer base, positions it as one of the more immediate markets for Lobster.cash adoption, particularly for automated B2B procurement and agent-to-agent marketplace payments. For developers in South Asia, Crossmint's single API covering both card networks and USDC lowers the barrier to building agentic payment applications without requiring a proprietary token or significant crypto market exposure.

Africa's situation is more layered. The continent handles roughly 70% of the world's $1 trillion mobile money market, but Visa and Mastercard card penetration remains far lower than in Asia or Europe. This limits the direct utility of Lobster.cash's card rails for most African users in the near term. The stablecoin rail is a more practical starting point. Nigeria accounts for approximately 40% of Sub-Saharan Africa's stablecoin inflows, and Kenya ranks fifth globally for transactional stablecoin use. For African freelancers and B2B exporters paying international platforms for cloud compute or SaaS subscriptions, agent-managed USDC payments are the more accessible entry point. Regulatory frameworks in both countries have advanced recently, with Kenya's VASP Act and Nigeria's Investment and Securities Act (2025) both taking effect in 2025, but neither addresses the question of legal accountability when an autonomous agent executes an unauthorized transaction.

What comes next

Mastercard has already expanded live agentic transactions to Europe (via Banco Santander in March 2026), Latin America and the Caribbean, Singapore, Malaysia, and South Korea. Wider deployment through payment processors including Stripe, Adyen, and Checkout.com is expected from 2027 onward. The competing x402 protocol, backed by Coinbase, has processed more than $600 million in transaction volume and supports around 500,000 active AI wallets. Two additional frameworks are shaping the competitive landscape: Visa TAP (Trusted Agent Protocol), which focuses on agent legitimacy through signing keys and delegation tokens and serves as a direct architectural rival to Mastercard's Verifiable Intent standard, and Google's Agent Payments Protocol (AP2), with which Mastercard's Verifiable Intent is designed to integrate. Taken together, these overlapping standards signal that the infrastructure race for autonomous agent payments is already well underway. The question of legal liability when an autonomous agent executes an unauthorized transaction remains unresolved globally, and card networks, regulators, and developers will need to address it as transaction volumes grow. McKinsey projects that AI-mediated commerce could reach $3 to $5 trillion by 2030, a figure that gives context to why card networks are moving to embed governance frameworks now rather than after the transaction volumes arrive.