$48M+ · AI + Banking Infrastructure

What they're building

Every payment rail built today assumes a human at one end of the transaction. ACH takes three days. Credit card fees run at 3%. KYC flows require a face. Fraud detection flags automated behaviour. None of that was designed for a world where 80% of economic activity is initiated by an AI agent rather than a person. Catena is building the financial institution that was designed from scratch for that world. The platform has three layers. The identity layer gives AI agents cryptographic identities, reputation scores, and the ability to authenticate themselves to counterparties without requiring a human to be in the loop. The payment layer uses stablecoins (primarily USDC, which Sean Neville co-invented) for near-instant, low-cost settlement that agents can execute programmatically without the delays and fees of legacy rails. The compliance layer handles KYC, AML, and risk management for autonomous actors rather than individual humans, a genuinely novel regulatory design problem. The Agent Commerce Kit (ACK) is the open-source protocol layer released at the seed announcement: building blocks for agent identity, payment, and commerce that any developer can use independently of Catena's licensed financial services. It is Catena's bid to set the standards for the category rather than just build a product in it. Catena is also the first company to apply for a US OCC national trust bank charter specifically for AI agent financial services, seeking regulatory authority to hold customer funds and process payments for agent transactions. The charter application is itself a moat: the compliance infrastructure required to get there is not something a new entrant can replicate quickly. The platform stack is Python and TypeScript, deployed on AWS and CloudFlare, using multiple LLM providers, and integrating with emerging agent protocols: A2A, x402, MCP, and LangGraph.

Why this matters

The agent economy is not a future scenario. Coinbase launched the first wallet purpose-built for autonomous AI agents in February 2026. OKX debuted an agent payments protocol in April. Deloitte has forecast that AI and stablecoins will reshape financial services by 2030. The infrastructure race is happening now, and the category is still wide open. The OCC charter application is the competitive moat argument. Most agentic payments companies are building at the application layer: they are connecting agents to existing rails, wrapping existing banks, and hoping the compliance complexity does not catch up with them. Catena is applying to become the bank. That is a fundamentally different product: it does not depend on underlying institutions that were not designed for agents, it cannot be pulled by a banking partner who decides the risk profile is unacceptable, and it can design its own risk and compliance systems specifically for autonomous actors. If the charter is granted, Catena holds customer funds directly and processes payments under federal authority. No other company in the agentic finance space is attempting this. A16z crypto's Chris Dixon named the investment thesis precisely at the seed: "AI agents need bank accounts and payment rails they can access programmatically, along with clear rules and controls that keep humans in charge." The Series A validation from Acrew Capital, General Catalyst, and QED arriving alongside a16z's continued support confirms the thesis has only strengthened in the twelve months since.