Possible Should we give AI a bank account?
30 snips
Apr 1, 2026 Sean Neville, co-founder of USDC and now leading Katana Labs to build AI-native banking, discusses agent-native financial infrastructure. He covers why current rails fail for autonomous agents, how stablecoins and Know Your Agent could enable machine-to-machine commerce, and the need for standards, deterministic guardrails, and compliance-first design as agent finance scales.
AI Snips
Chapters
Books
Transcript
Episode notes
Enterprise Back Office Will Lead Agent Adoption
- Early, high-impact adoption of agent transactions is likelier in enterprise back offices (cash flow, reconciliation, compliance) than consumer storefronts.
- These automated workflows already embed LLMs and are lower-risk, high-value targets for agentic finance.
Automate Monitoring But Keep Human Oversight
- Automate transaction monitoring with AI agents but keep human compliance experts to orchestrate, review, and file SARs.
- Sean says agent-driven monitoring cuts false positives and changes humans' tasks from manual triage to orchestration.
Map Agents Back To Licensed Operators Today
- Near-term KYA maps agents back to licensed human or business operators using cryptographic credentials; long-term KYA may recognize government-issued agent identifiers.
- This lets banks verify permissioned actions even when interacting only with an agent's surface.
