Visa Crypto Labs has introduced a command line interface (CLI) tool that enables AI agents to autonomously execute on-chain financial transactions without human involvement at the moment of payment. Visa Crypto Labs operates as the research and innovation division within Visa, focusing on advancements in digital assets, blockchain, and emerging payment infrastructure. Its work centers on creating protocols and products to drive the practical integration of blockchain with mainstream payment systems.
How the Technology Works
The CLI is tailored for developers building on large language models and AI architectures, allowing these agents to control their own blockchain wallets. These wallets can be loaded with stablecoins such as USDC or other digital assets, which the AI can then use to pay for resources like API keys, cloud computing power, or data sets. When the agent identifies a need, it can initiate and complete a payment autonomously, enabling operations to continue uninterrupted by manual approval workflows.
This operational model uses account abstraction and smart contracts to build programmable wallet infrastructure for AI bots directly on blockchain networks. Account abstraction lets wallets process transactions based on programmable rules instead of requiring a private key signature for each spend. This enables automated, rule-based payments that were previously difficult to support with conventional wallet structures.
Policy Controls and Developer Integration
Despite the potential for autonomous execution, Visa has incorporated a policy engine within the CLI to ensure oversight. Human owners can preset spending caps, designate approved merchants through whitelisting, and set expiration limits for the agent’s accessible budget. The AI operates independently only within these programmed boundaries, providing a balance between automation and control.
The decision to offer a CLI rather than an API is designed to serve the environments where most AI agents are created, tested, and deployed. Developers frequently work within terminal-based settings, and a CLI integrates naturally into those workflows. By skipping the additional layer of abstraction that comes with an API, Visa aims to speed adoption and lower the friction for integration with AI systems.
Market Context and Strategic Direction
The push toward autonomous transactions is supported by a recent joint report from Visa and Coinbase, which projects that agentic spending could generate over $500 billion in global transaction volume by 2027. While directional, this estimate signals Visa’s commercial intentions and the broader growth of automated payment flows powered by machine intelligence.
The company signaled further commitment to this strategy with a partnership announced on March 14 with Sky, previously known as MakerDAO, aimed at incorporating the USDS stablecoin into the new payment framework. This links Visa’s technical infrastructure directly with a liquid, widely recognized on-chain stablecoin, strengthening its ecosystem for autonomous agent spending.
The trend of AI agents procuring digital resources—such as data, compute, and APIs—is already underway. Until now, these transactions have relied largely on traditional payment cards tied to developer accounts with human intervention. Visa’s new CLI is intended to automate this process at scale, reducing human touchpoints and enabling seamless machine-to-machine commerce.
The actual market scale for autonomous agent payments remains to be seen, but Visa’s ongoing technological buildout is designed to support and potentially capture a substantial share of this emerging infrastructure layer.
