AI broke the economics of software. Vayu wants to be the first to tell a CFO which customers are losing them money

Here is a problem the software industry did not have three years ago.

A customer signs a contract for an AI product. They use it heavily, which is what every vendor wants. Every one of those uses costs the vendor money: tokens, compute, third-party model calls. The invoice goes out at the agreed price. The cost of delivering what was invoiced is somewhere in a cloud bill nobody has matched to that customer. And at quarter-end, finance discovers that the best customer on the revenue chart was a loss on the margin one.

“AI broke the old math of software,” says Erez Agmon, CEO and co-founder of Vayu. “A SaaS seat used to cost almost nothing to serve. Today a single heavy customer can wipe out the margin on an entire contract, and most finance teams find out a quarter later.”

Vayu , an AI revenue management platform used by finance teams at Vi, Groundcover, Narmi, Simetrik, Dataplor and Aquant, launched its Revenue Intelligence Hub, an insights and margin layer that sits on top of the company’s metering infrastructure and revenue agents and shows a CFO, customer by customer, whether pricing is keeping up with the cost of delivering AI. Alongside it, the company published the CFO Signal Report 2026 with PwC and The SaaS CFO on how usage-based, hybrid and AI pricing are reshaping revenue operations.

The first is that AI products carry variable cost . Every action a customer takes consumes something: a model call, a GPU minute, an embedding, a third-party API. Vendors have responded by moving pricing to usage , credits and outcomes, so that revenue tracks consumption.

The second is newer. A wave of AI-powered services companies now sell outcomes instead of hours , with contracts built on milestones, step-up commitments, ramp schedules and success-based fees. What was promised, what was delivered, what can be billed this month and what it cost to deliver it are now four different numbers in four different places.

“Pricing and cost now move together, every day, and the revenue engine has to see both,” he says.

Vayu’s answer is organised in three layers, and the new Hub is the top one.

Infrastructure. A metering and revenue data layer that gives engineering a single integration point for usage data. It ingests and validates up to one million billing events a day and unifies them with contracts, CRM and billing data from systems including Snowflake, AWS, Azure, Salesforce, Stripe and NetSuite. Developers connect once; after that, a pricing change does not need an engineering ticket.

Agents. On top of the data, AI agents run the revenue lifecycle end to end. Contract extraction turns agreements, including milestones, step-ups, ramps and outcome-based terms, into billing and pricing logic. Billing execution, ASC 606-ready revenue recognition, reporting, and agent-led collections and reconciliation follow. Routine work runs automatically; finance handles exceptions.

Intelligence. The Revenue Intelligence Hub’s Insights Agent answers plain-language questions about usage, contracts, pricing, billing, revenue and margins, grounded in live data, and surfaces what finance would otherwise learn at month-end: customers whose consumption is outpacing their plan, pricing tiers where cost-to-serve is eroding margin, contracts leaking revenue between agreed and invoiced terms, renewals at risk.

The practical promise is that a CFO can ask “which customers are below margin this month” and get an answer today rather than a reconstruction in six weeks.

The report, produced with PwC and The SaaS CFO, is based on insights from roughly 100 CFOs and finance leaders at scaling B2B companies, and its title does not hedge: the revenue engine is hitting its breaking point.

The headline numbers describe a finance function absorbing a pricing shift by hand. Seventy-two percent of CFOs say billing complexity is their number one operational friction point, ahead of headcount, compliance and the close process. Fifty-eight percent are running hybrid pricing models, seat plus usage plus one-time fees, with no unified billing layer, managed across three or more systems. Forty-one percent say revenue recognition is still partially manual, even with an ERP in place. And the teams that have automated their billing infrastructure close the books twice as fast as those on legacy or manual workflows.

The report’s framing is that most SaaS billing infrastructure was built for a business that no longer exists, and that as pricing moves to usage-based, hybrid and seat-plus-consumption models, the gap between where CFOs want to be and what their tools allow is widening. Inside it: billing-stack benchmarks and what the top quartile looks like, where ASC 606 compliance is breaking down, how CFOs are trying to forecast revenue on variable contracts, the order in which modernised teams upgraded, and a CFO Confidence Index on readiness for board-level revenue scrutiny. It is available free at signal-report.withvayu.com .

“Pricing has become a growth lever, and finance has been handed the complexity without the tools to manage it,” Agmon says. “The companies that win in AI will be the ones that know their unit economics in real time, not at quarter-end.”

Agmon’s background explains the bet. He spent a decade across fintech operations and early-stage investing, including as a founding team member at PayEm, part of the wave that automated accounts payable. Receivables, the money coming in, stayed in spreadsheets, because every customer’s pricing was different and every pricing change meant an engineering ticket. Vayu, founded in 2024, was built to be the second wave: finance-owned revenue, from contract to cash.

The customer evidence so far is about speed and error rates. Vayu reports 75 percent faster billing, 90 percent faster reconciliation and zero spreadsheet errors across its base, with customers typically live in 14 days or less. Narmi’s VP of finance and accounting, Brian Aber, described the company’s old process as “the world’s busiest spreadsheet” and its billing today as “an automated process we can trust.” Simetrik’s CFO, Felipe Pardo, said Vayu lets the company “see customer consumption as it happens” and handle renewals and billing “before they become month-end issues.”

Vayu is SOC 2 certified and Built for NetSuite. It has raised millions of dollars from Flint Capital and The Garage, with participation from the founders of Melio and managing partners of SoftBank’s previous funds, and is headquartered in New York with R&D in Tel Aviv.

The claim that matters is the margin, and it is the hardest to deliver. Showing cost-to-serve per customer requires matching cloud and model spend to individual accounts, which means Vayu’s infrastructure layer has to see the cost side as cleanly as it sees the usage side. The company says it does; the CFOs who adopt the Hub will be the ones to confirm it.

The second question is whether finance teams, who have spent a decade learning to distrust automation around money, will let agents run billing, recognition and collections. Vayu’s design keeps people on the exceptions, which is the right posture, and its customer results suggest the routine side holds up.

But the underlying problem is not going away. AI made every customer a variable cost, and outcome-based contracts made every invoice a reconciliation. Whoever shows finance both numbers, in the same place, on the same day, is going to have a lot of CFOs’ attention.

Contributed article. Not produced by the TNW newsroom and does not reflect the editorial stance of TNW.