Crypto Firms Build AI Agent Payment Infrastructure
Cryptocurrency companies are developing payment infrastructure specifically for AI agents, according to a report from Tiger Research. The effort aims to enable autonomous agent transactions, programmable payment flows using smart contracts, and cross-chain interoperability. This new financial rail is being built to support an emerging economy where AI agents can initiate, verify, and settle payments without human intervention.
- Startups are emerging to provide the financial infrastructure for AI agents, including InFlow, founded by a former PayPal executive, and Natural, which raised $9.8 million in seed funding to focus on B2B and embedded agentic payments. Other companies in this space include Nevermined, Circuit, and Masumi, which are developing protocols for agent-to-agent transactions. - A key technical challenge is enabling privacy-preserving transactions between multiple agents without a centralized intermediary. A recently proposed protocol, PrivateX402, uses payment channels to allow a user to allocate budgets across many agents while hiding individual payment amounts from on-chain observers. - Open-source multi-agent orchestration frameworks are critical for development, with Microsoft's AutoGen and LangChain being prominent options. AutoGen excels at conversational, collaborative agent systems, while LangChain is better suited for creating deterministic, modular pipelines for tasks like retrieval-augmented generation. Newer frameworks like CrewAI offer a role-based abstraction for agent coordination. - In China, the AI agent ecosystem is rapidly developing, with tech giants like Tencent (Hunyuan), Baidu (ERNIE Bot), and Ant Group (Lingji) offering agent development platforms. Startups like Butterfly Effect, with its "Manus" agent, have gained significant traction, signaling a shift toward agents that can perform complex tasks independently. - Recent research in AI agent architecture focuses on enabling agents to evolve and learn from experience. Papers such as "Self-Consolidation for Self-Evolving Agents" and "Agentic Memory" explore methods for agents to autonomously improve their skills and manage long-term and short-term memory. - The convergence of AI and crypto is seen by investors like a16z as requiring three core pillars: collaborative infrastructure for interoperability, verifiable identity for agents (Know Your Agent), and micropayment channels for real-time, usage-based transactions. Blockchain technology is positioned to provide these foundational layers, enabling agents to operate as autonomous economic actors. - On the regulatory front in China, there is no single comprehensive AI law, but a series of measures from bodies like the Cyberspace Administration of China (CAC) govern generative AI. Recent updates to crypto regulations in February 2026 reiterated a strict ban on most virtual currencies but created a narrow, state-controlled path for the tokenization of real-world assets (RWA).
