Building the Agent Economy for Digital Assets

At its Real Up summit in Cannes, Ant Group introduced Anvita, a platform designed to enable artificial intelligence agents to autonomously manage and execute cryptocurrency transactions. This marks a significant step toward what executives envision as an “agent-to-agent” economy where bots can hold assets and make payments with minimal human intervention.

Anvita launches with two key components:

  • Anvita TaaS (Tokenization-as-a-Service): Provides institutions with custody, treasury tools, and infrastructure for tokenizing real-world assets (RWAs).
  • Anvita Flow: A platform enabling AI agents to connect, coordinate tasks, and settle payments in real time.

“Pure RWA is just the ‘static infrastructure’ of digital assets,” said Zhuoqun Bian, president of blockchain business at Ant Digital Technologies. “The real transformation lies in moving toward an onchain agentic economy, where autonomous agents will not just analyze data — they will hold assets, execute trades, and optimize portfolios.”

The Race to Enable AI Commerce

Ant Group’s announcement comes amid broader industry efforts to build infrastructure for AI-driven commerce. Competitors include:

  • Visa: Launched the Trusted Agent Protocol for card-rail checkout
  • Coinbase: Developed x402 for stablecoin micropayments
  • Google: Introduced the Agent Payments Protocol
  • Mastercard: Acquired BVNK, a stablecoin infrastructure firm, in a $1.8 billion deal (the largest ever)

This competition suggests that traditional payment networks also see blockchain settlement as playing an increasingly important role.

Addressing Adoption Hurdles

While interest in digital assets remains high, adoption has been hindered by usability challenges and governance concerns. As Mesh CEO Bam Azizi noted, “The biggest problem in crypto is not adoption; it’s the user experience.” Simplifying payments to the point where they can be used by anyone—even those unfamiliar with blockchain technology—will be crucial for broader acceptance.