Ant Group Launches Anvita: The Platform Where AI Agents Buy, Sell and Pay With Crypto on Their Own

For most of the internet’s history, every financial transaction — every payment, every trade, every transfer — started with a human making a decision. That is about to change.

On March 31, 2026, Ant Digital Technologies, the blockchain division of Ant Group, unveiled Anvita at the company’s Real Up summit in Cannes, France. Anvita is a platform designed specifically to let artificial intelligence agents — not humans — become the primary participants in financial transactions. These agents can hold assets, execute trades, and make payments instantly using stablecoins, with little to no human involvement at any step.

Ant Group is the parent company of Alipay, one of the world’s largest digital payment networks. When a company of that scale makes a bet on AI-to-AI commerce, it is not a research experiment. It is a signal about where the entire global payments industry is heading.

What Is Anvita and What Did Ant Group Actually Launch?

Anvita is described by Ant Digital Technologies as infrastructure for an “agent-to-agent economy.” The idea is straightforward: just as the internet made it possible for humans to transact with each other without physical presence, Anvita makes it possible for AI systems to transact with each other without any human pressing a button.

The platform launched with two core products.

The first is Anvita TaaS — Tokenization-as-a-Service. This product is aimed at institutions and allows them to convert real-world assets into blockchain-based digital tokens. It includes tools for asset issuance, secure custody, and treasury management. In simple terms, a bank or financial institution can use Anvita TaaS to put a bond, a fund, or any other financial instrument onto a blockchain — making it faster to transfer, easier to divide, and more transparent to audit.

The second is Anvita Flow. This is the more groundbreaking product. Anvita Flow is a live network where AI agents can register themselves, discover other agents, coordinate tasks between each other, and settle payments in real time. An AI agent managing a trading portfolio could use Anvita Flow to find a data provider agent, purchase market data instantly in fractions of a cent, and execute a trade — all without a human approving any individual step.

Also Read: SEC Approves Nasdaq Rule to Trade Tokenized Securities: Wall Street Meets Blockchain

What Technology Powers Anvita Flow?

The payment layer inside Anvita Flow is built on the x402 protocol, a standard developed jointly by Coinbase and Cloudflare. x402 allows stablecoin payments to be made directly over HTTP — the same protocol that powers every website on the internet. In practical terms, this means an AI agent can pay for a service, a data feed, or a compute task the same way a browser loads a webpage: instantly, automatically, and without any billing system, invoice, or subscription model in between.

Payments on Anvita Flow settle in USDC, the US dollar-pegged stablecoin issued by Circle. Transactions can be completed in fractions of a cent, making it economically viable for agents to pay each other for tiny, high-frequency tasks — something traditional payment rails built around credit cards and bank transfers cannot support.

The platform also includes an Agent Store for developers. This marketplace contains pre-built modules covering data collection, financial analysis, portfolio management, and more. Developers can add new agents to the store, and the infrastructure is compatible with existing AI frameworks including Claude Code, making it accessible to a wide range of engineering teams.

Why Did Ant Group Build This? What Problem Does It Solve?

To understand why Anvita matters, it helps to understand what is broken in the current system.

Today, when a software application needs to pay for an external service — a data feed, an API call, a cloud compute job — it goes through a billing system. That system typically involves invoices, subscriptions, monthly charges, and human approval somewhere in the process. For a single payment, this is manageable. For an AI agent making thousands of micro-decisions per second, each of which might require paying another system for data or compute, this model completely breaks down.

Zhuoqun Bian, President of blockchain business at Ant Digital Technologies, described the problem precisely at the Cannes summit. He said that pure real-world asset tokenization is only the static infrastructure of digital assets. The real transformation, he argued, lies in moving toward an on-chain agentic economy — one where autonomous agents do not just analyse data but actively hold assets, execute trades, and optimise portfolios on their own.

Anvita is Ant’s answer to that problem. It creates the plumbing that AI agents need to operate as economic participants — not just as tools that humans direct, but as independent actors that can earn, spend, and transact within a defined set of rules.

Also Read: AI in Smart Contract Auditing Explained

Who Else Is Building in This Space?

Ant Group is not alone. The race to build payment infrastructure for AI agents has attracted some of the largest companies in both traditional finance and crypto.

Visa has launched its Trusted Agent Protocol, focused on enabling AI agents to complete purchases through existing card payment rails. The approach is different from Anvita — Visa is working within the traditional card infrastructure rather than building on stablecoins — but the goal is the same: let AI agents buy things without a human approving each transaction.

Coinbase developed the x402 protocol — the same payment standard now powering Anvita Flow — and has built its own Agentic Wallets product, which allows AI systems to hold crypto balances and complete transactions on-chain without human approval.

Google launched its Agent-to-Agent (A2A) Protocol in September 2025, now supported by more than 60 organisations, focused on how AI agents from different companies can discover each other and work together.

Mastercard acquired stablecoin infrastructure firm BVNK for 1.8 billion dollars — the largest stablecoin infrastructure acquisition ever recorded — signalling that traditional payment networks also see blockchain settlement as central to their future.

What Does This Mean for Businesses Building on Blockchain?

For teams building blockchain development solutions, the Anvita launch is one of the clearest signals yet that the next generation of blockchain applications will not be built around human users clicking buttons. They will be built around AI agents making decisions autonomously.

This changes what good blockchain infrastructure looks like. An AI agent operating on Anvita Flow needs a smart contract that can be called programmatically at high speed, with predictable gas costs, clear access controls, and audit trails that satisfy both technical and regulatory requirements. It needs a token standard that supports sub-cent transactions without friction. It needs custody infrastructure that does not require a human to sign off on every individual payment.

For developers building AI-powered applications, Anvita represents the moment when AI and blockchain stop being two separate conversations. If your AI application needs to pay for data, compute, or services — and most production AI applications do — you are now in the blockchain space whether you planned to be or not.

The commercial implications are also significant. According to analysis by McKinsey, AI-powered software could drive between three and five trillion dollars in global consumer spending by 2030. If even a fraction of that activity flows through agent-to-agent payment rails like Anvita Flow, the volume of on-chain transactions will dwarf anything the crypto industry has seen before.

Is Anyone Actually Using It Yet?

Honestly — not at scale. This is important context.

Current transaction volumes on the x402 protocol — the payment standard Anvita Flow uses — are still very low. Analysis has found that a significant portion of current activity is test transactions rather than real commercial usage. The Solana Foundation reported over 15 million agent transactions on its network, but agent-to-agent commerce at meaningful economic scale does not yet exist in a mainstream sense.

What Anvita represents is infrastructure being laid ahead of demand. The same thing happened with the internet in 1996, with mobile payments in 2010, and with cloud computing in 2012. The companies that built for those transitions early captured the market when volume arrived. The companies that waited found the market already claimed.

Ant Group, Visa, Coinbase, Google, and Mastercard are not building agent payment infrastructure because it is profitable today. They are building it because they believe it will be the dominant model within three to five years — and none of them can afford to be second.

The Bottom Line

Ant Group’s Anvita launch is the clearest signal yet that the next phase of the internet — one where AI agents transact directly with each other, without human middlemen, using blockchain rails and stablecoins as the settlement layer — is no longer a theory. It is now a product you can build on.

For every business operating at the intersection of AI and blockchain, that is the most important development of 2026 so far. The infrastructure is being built. The protocols are being written. The question is not whether agent-to-agent commerce will happen. It is who will build the applications that run on top of it when it does.