In the early 1990s, there were serious, well-funded proposals for proprietary internet infrastructure. CompuServe, AOL, and Prodigy each offered closed networks where content and commerce would flow through a single corporate gatekeeper. They were convenient, curated, and technically functional. They were also the wrong model. We got TCP/IP and HTTP instead — open protocols that no single entity controlled, where the network was composable by design. The open standard won decisively, and the internet became the most powerful economic infrastructure in human history.
Agentic finance faces the same fundamental choice, right now, mostly implicitly, in the architectural decisions being made by infrastructure providers, enterprises, and developers. The outcome of those decisions will determine whether the agent economy is genuinely open — interoperable, composable, permission-free — or owned by a small number of platforms that control access to the financial rails that agents depend on.
This isn't an ideological argument. It's an engineering and economic argument. Open standards produce better outcomes for everyone except the entity that would otherwise extract rent from a closed system.
How AI agents use money — and why the architecture matters
To understand why open standards matter for agent finance specifically, you need to understand how agents use money differently from humans. A human makes payments occasionally, interactively, with meaningful judgment at each transaction. An agent makes payments programmatically, at high frequency, with structured logic rather than human discretion.
An agent managing procurement might evaluate 500 invoices per day and execute payment decisions against a policy engine without a human in the loop. An agent coordinating a research task might pay for data API access, compute resources, and sub-agent services as part of a single autonomous workflow. An agent managing a corporate treasury might rebalance across multiple protocols in response to market conditions while simultaneously managing liquidity for operational payments.
In each of these scenarios, the agent doesn't care about the identity of the settlement provider — it cares about whether the payment goes through, at what cost, with what finality, and with what auditability. Those requirements point to open settlement infrastructure, not closed platforms.
The proprietary model — what it offers and what it costs
The proprietary model for agent finance is vertically integrated: wallets, policy enforcement, payment routing, settlement, and identity all through a single provider. It has real advantages for early-stage development. You integrate one SDK, you have one support relationship, the parts are designed to work together.
The costs emerge at scale, and they're structural rather than incidental:
- Network isolation. Your agents can transact with other agents on the same platform. Agents on different platforms require custom integrations or simply can't transact at all.
- Fee leverage. When the provider controls the entire stack, fees are whatever the provider decides. You have no alternative — switching costs are enormous once you've built a production system around a proprietary API.
- Operational dependency. If the provider changes terms, pauses operations, or exits the market, your agents stop. That's a single point of failure at the infrastructure level that no enterprise risk framework should accept.
- Regulatory concentration. A single custodial provider accumulating agent finance at scale becomes a systemically important institution with regulatory obligations that may not serve your interests.
The open standards model — what it requires and what it returns
The open standards model builds on protocols rather than platforms. x402 is the payment negotiation protocol — HTTP-native, requiring no registration, compatible with any agent framework and any API server. USDC settles on public blockchains with on-chain finality and auditability. Agent identity frameworks use verifiable credentials that any verifier can check without a central registry.
Building on open standards requires more initial work: you're composing infrastructure rather than calling a single SDK. But the returns compound over time:
- Interoperability by default. Any x402-compatible agent can transact with any x402-compatible service, regardless of which infrastructure provider either side uses.
- Fee competition. When no single provider controls the stack, providers compete on price. Over time, fees approach marginal cost.
- Composability. Open blockchain settlement means your agents can compose with the entire on-chain financial stack — yield protocols, lending markets, treasury management — without custom integrations.
- Resilience. No single provider controls your agents' financial access. If one provider fails, you switch without losing access to funds or counterparty relationships.
The network effect argument
The most important structural advantage of open standards is network effects. Every API server that implements x402 becomes reachable by every x402-compatible agent. Every agent framework that supports x402 payment handling expands the universe of services that agent can access. The network grows with each participant, and no single entity captures the value of that growth.
Compare this to the proprietary model, where network effects benefit the platform. The larger the closed network gets, the stronger the lock-in for everyone in it. Value accrues to the platform, not the participants. This is the AOL model — briefly dominant, ultimately superseded by the open web.
x402 already has 35 million transactions in production, with backing from Coinbase, Google, Anthropic, Visa, AWS, and Cloudflare. That's not a bet on a new standard — that's the early signal of a standard that has already found product-market fit.
What open standards mean in practice for builders
Building on open standards doesn't mean building everything yourself. It means choosing infrastructure providers that implement open protocols rather than proprietary ones. An x402-compatible wallet provider, USDC settlement on Base, and open credential frameworks for agent identity — these can all be assembled from competing providers at each layer.
Proco is built on this model. x402-compatible out of the box. USDC settlement. Non-custodial wallet architecture where your agents hold their own keys. Open KYA credential frameworks that any verifier can check. We compete on tooling, on developer experience, on the quality of our policy engine — not on lock-in.
The alternative — a closed ecosystem where agent finance only works if everyone uses the same provider — is the wrong bet for the same reason the closed internet proposals of the 1990s were wrong. We've seen how that story ends. The open standard wins. We're building the open standard infrastructure for agent finance.
Further reading: Open Infrastructure vs Walled Gardens · Building on x402