Original title: Agent-to-Agent Finance: Blockchain Payments and Trust Infrastructure for Autonomous AI Agents
Hui Gong · June 2026
Computer programs powered by artificial intelligence can now interpret goals, use digital tools, negotiate, and send payments or blockchain transactions. This creates a big challenge: when software acts as an economic partner, financial markets need reliable ways to verify identities, authorize deals, process payments, track reputations, and hold systems accountable. To solve this, the researchers introduced the concept of agent-to-agent finance. They examined how technologies like programmable settlementThe ability to automatically execute payments when certain conditions are met, using code., smart wallets, decentralized registries, and verifiable computationA way to prove that a computer program ran correctly without having to re-run it. can reduce coordination friction between software agents. These systems let machines discover other agents, buy digital services, pay each other, and create clear audit trails. The authors conclude that the primary design goal is bounded autonomyGiving agents freedom to act but within limits to prevent harm or chaos.: markets must let independent software trade without becoming fragile, hidden, or unaccountable.
You are at a farmer's market where you need to trust vendors, pay, and keep records. If everyone started trading without IDs or payment systems, it would be chaotic. In this paper, AI agents face the same problem when transacting autonomously. Here, a system provides the identities, wallets, and transaction records that keep the market orderly.
AI agents are starting to transact autonomously, but current financial systems lack the infrastructure for agent identity, payment, and accountability. Without new systems, agent-driven markets could become opaque and fragile. This matters for financial institutions, regulators, and anyone relying on AI for economic decisions.
The paper is conceptual and does not present a full implementation or empirical evaluation. It focuses on identifying the problem and proposing a framework, but does not solve all challenges of bounded autonomy. The reliance on blockchain may not be suitable for all financial contexts, as the authors acknowledge.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Agent-to-Agent Finance: Blockchain Payments and Trust Infrastructure for Autonomous AI Agents — Hui Gong
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.