According to Forbes, a new startup called InFlow is launching out of stealth today, founded by former PayPal executive Jim Nguyen. The company’s goal is to build an autonomous payment system specifically for AI agents, enabling them to transact, hire other agents, and pay for services. Nguyen, whose previous startup RubyCoins was acquired by PayPal, spent six years there as head of global business development. He frames this as the next evolution: PayPal for the web, Apple Pay for apps, and now InFlow for AI agents. The core problem they’re tackling is that while agents can automate complex digital tasks, they currently can’t perform the basic commercial act of signing up for a service and paying for it.
The Agent Economy Needs A Wallet
Here’s the thing: Nguyen’s pitch makes a ton of sense if you buy into the agent future. Right now, an AI can write the code for a website, but then it hits a wall. It can’t go get a domain name, provision hosting, or hook up a database because all of that requires a credit card and a sign-up flow. Basically, the agent does the “smart” work and then hands the boring, tedious “plumbing” back to a human. InFlow wants to embed that payment and onboarding capability directly into the agent’s workflow. So the agent could, within its development environment, create an account, pay for a database, and spin it up—all without you lifting a finger. That’s the friction they’re aiming to vaporize.
Budgets, Accountability, and Rogue Agents
Now, the immediate question anyone would have is: are we just going to give AI agents unlimited spending power? That seems… unwise. InFlow says agents will operate with defined budgets, much like a human assistant or a department would. They’ll also be formally tied to the human who instantiated them, creating a chain of accountability. The idea is to prevent rogue agents from going on a cloud services shopping spree. But let’s be real, the security and oversight here will be everything. It’s one thing for a human to overspend; it’s another for a misconfigured agent to autonomously rack up thousands in API calls before anyone notices.
A Fundamental Shift In Who (Or What) Is A Customer
This is where the philosophy gets interesting. Nguyen is arguing that AI agents represent a new class of customer for service providers. They’re a “buying customer with earning capabilities.” He brushes off pushback from companies like Amazon (which told Perplexity to keep its agents off Amazon.com) as a temporary issue. The bet is that once agents are reliable economic actors, businesses will want their money. It speaks to a hybrid ecosystem we’re just starting to see, where digital and human workers mix. As OpenText’s CTO noted, we might even need HR systems for agents—onboarding, paying, and retiring them. If that future arrives, you absolutely need a payment rail built for it. InFlow is trying to be that rail.
Is The Timing Right?
So, is this genius or wildly premature? Dozens of major platforms from Adobe to Salesforce are building agentic capabilities, but we’re still in the early, often clunky, stages of agent technology. Building the payment system before the economy fully exists is a classic infrastructure play. It’s risky, but if it takes off, InFlow could become fundamental plumbing. The model will likely be transactional, taking a fee when an agent spends. The beneficiaries? Every company building agents that need to interact with the real commercial world. And, of course, the service providers who get paid. It’s a big vision. Whether the world is ready for AI agents with credit cards is the multi-billion dollar question.
