A monetization platform for the software that works instead of the software you log into. Design the pricing, track the margin, hand the customer a receipt.
There is a genre of startup that does not want to be famous. It wants to be load-bearing. Stripe wanted to be the thing under the checkout button. Twilio wanted to be the thing under the text message. Paid wants to be the thing under the invoice that an AI agent sends you - the small, unglamorous, absolutely unavoidable moment where an autonomous piece of software says "you owe me this much, and here is exactly why."
The interesting part is what Paid is not. It does not build AI agents. It has, as far as anyone can tell, no intention of building AI agents. In a market where roughly every company with a GPU is racing to ship an agent that books your travel or closes your tickets, Paid looked at the gold rush and decided to sell a very specific shovel: the cash register. Its founders are betting that the agents will get built by everyone else, and that everyone else will hit the same wall around month six - how do we actually charge for this thing?
The old answer was the seat. You sold software by how many humans logged in. That model held up for fifteen years and made a lot of people rich. But an AI agent does not log in. It does not occupy a seat. It does work - it saves a customer thirty points of margin, or resolves four hundred tickets, or generates a pipeline of leads - and none of that maps cleanly onto "how many people are using this." Paid's whole thesis fits into that gap. If the agent produces an outcome, charge for the outcome.
Our mission is to grow the AI agent economy by helping builders get paid for their agents.
Saying "charge for outcomes" is easy. Building the machinery underneath it is not, which is roughly why Paid exists. To bill for an outcome you first have to see the outcome - every action the agent takes, in real time - then translate those actions into a business signal, price it, and produce something the customer will actually pay against without arguing. Here is how the pieces fit.
Configure credits, usage-based, outcome-based, success fees, revenue share, or hybrid models - no custom billing code required.
Every invoice ships with quantified proof: time saved, cost cut, ROI delivered. The customer sees what the agent earned them.
Real-time revenue, cost, and profit for every customer, product, and individual agent action - plus scenario planning.
OpenTelemetry-based SDKs track each agent action live and translate it into a billable business signal.
Automated invoice generation tied directly to the value receipt - the bill and the proof arrive together.
Most billing platforms were built to meter API calls. Paid's argument is that metering is the least interesting way to price an agent - because the customer doesn't care how many calls happened, they care what got done. These are the levers Paid gives builders, from the familiar to the ambitious.
Bar length is editorial, not empirical - a read on where Paid's thesis leans, not a benchmark.
Manny Medina did not need to do this. He founded Outreach in 2011 and grew it, by the accounts he repeats, from an idea to a reported $4.4 billion valuation and around $250 million in recurring revenue before stepping back from the CEO role. That is the sort of resume that lets a person spend a decade advising, angel-investing, and answering email from a beach. Instead he picked invoicing.
That choice tells you something about how Medina reads the market. The flashy problem in AI is capability - can the agent do the job. The durable problem, the one that will still matter in five years, is commerce - how does anyone get paid when it does. He surrounded himself with people who have scar tissue on exactly that: a co-founder who built billing at Salesforce, one from early Palantir, one from early Pleo. This is a team that has, collectively, watched billing systems break at scale and lived to build the next one.
Founder of Outreach; grew it to a reported $4.4B valuation.
Serial founder; previously built billing at Salesforce.
Early Palantir.
Early Pleo.
Investors do not usually pour eight figures into a billing startup this fast. That they did is a tell: the smart money believes the agent economy is real, and that whoever owns the payment layer owns something structural. Lightspeed framed it as the agent economy's "$19 trillion problem." Paid is selling the answer to that problem's most boring, most necessary question.
Paid points to a roster of agent builders and SaaS companies moving off seat-based pricing. It also makes a specific, checkable claim: early adopters seeing 20-40% revenue growth within six months. Numbers like that are exactly the kind a skeptic should poke at - but they are also exactly the kind that, if true, explain why a room full of VCs wrote checks this quickly.
The bet here is not that AI is impressive - everyone already agrees on that. The bet is narrower and, in a way, more contrarian: that the constraint on the agent economy is not intelligence but accounting. That value-based pricing, a consulting fantasy for decades because value was never measurable, finally becomes real the moment every agent action is logged. That the company which turns those logs into an invoice nobody argues with becomes, quietly, indispensable.
That could be wrong. Pricing infrastructure is a crowded, competitive space, and "outcome-based billing" is easier to demo than to reconcile at quarter's end. But it is the kind of wrong that is worth being, because the middle of a story - the plumbing, the ledger, the register - is historically where the durable companies get built. Paid is a very deliberate bet on the middle.