IMF official Dan Katz discusses AI's near-term potential to reduce payment costs and longer-term role of tokenized money for agent transactions.
AI & Agents ·
An IMF official discussed how artificial intelligence can lower payment processing expenses in the near term, while positioning tokenized money as infrastructure that may eventually facilitate transactions between autonomous software agents. The remarks situate AI's immediate utility in cost reduction within existing payment systems, contrasting with a longer-term vision where digital assets specifically designed for agent-to-agent settlements could play a supporting role.
AI agents are evolving from advisory tools into autonomous economic actors with on-chain identities and crypto payment capabilities. These systems range from copilots that surface recommendations alongside human operators to delegated agents acting within preset spending rules, to autonomous agents executing transactions with minimal human checkpoints. Examples include agents autonomously booking millions of hotel stays on blockchain networks and AI-powered virtual cards managing consumer purchases within pre-authorized budgets.
The infrastructure supporting agent payments remains in early stages. Several questions persist around liability when autonomous systems fail, the standardization of agent-to-agent transaction protocols, and whether existing payment rails will suffice or whether purpose-built tokenized systems will become necessary as agent activity scales.