Lobster.cash and Mastercard have announced plans to integrate Agent Pay and Verifiable Intent to enable secure, card-based AI agent transactions.
The integration connects network-backed payment infrastructure with a standards-based authorisation layer, extending trusted card payments into open agentic ecosystems. Mastercard cardholders using OpenClaw will be able to authorise AI agents to make purchases on their behalf. Each transaction will be governed by issuer controls, authenticated through Mastercard's network, and cryptographically linked to the user's explicit intent. Furthermore, the solution also incorporates Basis Theory as the agentic credential layer.
Cryptographic authorisation as a trust layer
A central component of the integration is Mastercard's Verifiable Intent framework, co-developed with Google and aligned with the Agent Payments Protocol (AP2) and Universal Commerce Protocol (UCP). The framework produces cryptographic, tamper-resistant records of user authorisation, allowing issuers, merchants, and platforms to independently verify that each transaction occurred within the scope explicitly approved by the user.
Verifiable Intent is designed to be protocol-agnostic, functioning as a shared source of truth across open agent ecosystems. Its inclusion in the lobster.cash integration addresses a key accountability gap in agentic commerce: establishing who authorised an agent, under what conditions, and within what boundaries.
Mastercard Agent Pay has already been deployed by a number of financial institutions globally, including Santander, Commonwealth Bank of Australia, DBS, and UOB. The arrangement with lobster.cash extends that infrastructure into developer-led, open-platform environments.
Expanding to agentic platforms
Lobster.cash currently provides payment infrastructure across several agent platforms, including OpenClaw, Claude Code, Devin, Hermes, and Zo Computer. Mastercard Agent Pay will initially be available to OpenClaw agents through lobster.cash, with plans for broader rollout to additional supported platforms.
The integration will launch via an early access programme before expanding to the wider OpenClaw ecosystem. It gives agent users programmatic control over spending parameters, including amount, merchant category, timing, and payment method, while also preserving the issuer oversight and consumer protections associated with standard card transactions.
The arrangement reflects a broader shift in payments infrastructure, as network operators, issuers, and developers work to establish clear authorisation standards for autonomous agents operating across multi-platform environments. Through the process of anchoring agentic transactions to existing card rails rather than requiring proprietary wallets or separate credentialling systems, the integration lowers the barrier to adoption for both consumers and developers.
