Mastercard has launched Agent Pay for Machines (AP4M), a dedicated financial protocol that allows autonomous AI agents to execute high-frequency transactions without real-time human approval. The system features over 30 initial industry partners, including major crypto platforms Coinbase, RippleX, Aave Labs, OKX, Polygon, and the Solana Foundation. According to Mastercard's announcement, the framework addresses structural limitations in traditional payment rails by enabling high-frequency, low-latency microtransactions that typically trigger efficiency bottlenecks on legacy networks. The system integrates multi-rail settlement from day one, explicitly supporting regulated stablecoins alongside traditional bank accounts and card networks to power high-frequency machine commerce. At the center of the protocol is a multi-rail settlement system that incorporates stablecoin settlement directly into the software stack, allowing AI agents to utilize digitized assets to settle accounts instantaneously rather than routing exclusively through traditional credit lines.
The latest launch has expanded Mastercard's initial partner ecosystem significantly, with more than 30 organisations now part of the initial partner group working with Mastercard to validate priority use cases. The updated partner list includes Adyen, Ant International, BVNK, Checkout.com, Cloudflare, Coinbase, Getnet by Santander, Global Payments, Lovable Labs Incorporated, OKX, Stripe and Tempo among others. As reported by Mastercard, Agent Pay for Machines will create the conditions for a superbloom of AI business models. According to Mastercard chief product officer Jorn Lambert, machine payments can make it possible for services to be bought and sold among agents at fundamentally different scales than payments today — very high volumes, very small values, very fast and at extremely low latency. The service is positioned as different from point-of-sale or person-to-merchant payments, which are typically user-initiated and processed as discrete transactions.
Coinbase has officially launched 'Coinbase for Agents', a comprehensive platform that establishes dedicated accounts and sub-accounts specifically for AI agents to trade, manage money, and make payments on behalf of users. The platform provides agents access to Coinbase's advanced trading tools through natural language commands, enabling users to authorize tasks ranging from portfolio rebalancing to automated strategy execution. As reported by The Block, agents can conduct market research for users and be programmed to trade based on particular theses on active or passive basis. At launch, agents can trade spot cryptocurrencies, with equities planned for three weeks and prediction markets scheduled for early July. The system operates within user-defined spending and risk limits, with plans to open access to the entire Coinbase suite of products including derivatives markets. Coinbase has also developed tooling for agentic wallets through its Coinbase Developer Platform, indicating that wallet-level support for autonomous agents is becoming a competitive category. According to Coinbase CEO Brian Armstrong, users can now authorize AI agents to control their Coinbase accounts or sub-accounts, with the platform connecting large language models such as ChatGPT and Claude to Coinbase accounts.
Coinbase is integrating support for x402, an open machine-to-machine payments protocol developed at Coinbase, which allows agents to make small payments for services such as premium research, data APIs and computing resources without subscriptions or manual checkout processes. According to The Block, over 10,000 online merchants have integrated x402, providing an alternative route for AI agents to access paid tools and content on a per-use basis through a dedicated agentic app store. The protocol supports stablecoin alongside every major payment rail, including cards and ACH, enabling seamless machine-to-machine transactions across the platform. Coinbase has also integrated Coinbase Advisor into the platform, describing the service as an SEC- and CFTC-registered financial advisor that can provide recommendations and guidance directly within agent workflows. Meanwhile, Stripe has introduced payphone beta testing project, which allows AI agents to initiate payment requests through voice interfaces, and Projects service providing dedicated billing and subscription management infrastructure specifically designed for AI agents.
Mastercard's AP4M relies on four primary operational functions: credentialing, programmatic permissioning, transacting and settling. Users define spending parameters, authorization thresholds, and intent verification rules beforehand, with software agents issued restricted digital network tokens that dynamically enforce these operational limits. As reported by Cryptopress, the system prevents autonomous capital drain while verifying that consecutive background business transactions remain fully compliant. Coinbase has implemented comprehensive safety features allowing users to create isolated sandboxes for agents to operate within, limiting their access to main Coinbase accounts. Users can set parameters including transaction and withdrawal limits, maximum trade sizes, and restrictions on which services agents can access. MetaMask's controlled wallet approach addresses security gaps by enforcing guardrails by default, including restrictions on transfer amounts, whitelisted contract interactions, and caps on gas spending.
AI agents are rapidly moving from proof-of-concept stages toward practical commercial applications, with payment infrastructure emerging as a critical bottleneck in this transition. NousResearch's Hermes AI Agent has demonstrated impressive autonomous shopping and payment abilities, completing entire e-commerce transactions without human intervention. OpenAI is extending Codex Agent into financial banking, signaling that AI agents are entering enterprise-grade application scenarios with extremely high security and compliance requirements. Industry experts forecast that autonomous agents could account for as much as 20% of e-commerce activity by 2030. The market adoption timeline will likely follow a pattern similar to other fintech innovations: early adoption in tech-forward sectors, followed by gradual expansion into mainstream commerce, with consumer-facing applications seeing slower adoption due to trust and liability concerns while business-to-business scenarios advance more rapidly. Whether developers and AI teams build on MetaMask's approach at scale will depend on how effectively the permission model balances security with the flexibility agents need to operate across DeFi protocols. Following the announcement, COIN stock climbed 4.8% to $161.48, although shares remain down more than 25% over the past month.
