AI-Era Startups Inflate ARR With Pilot Deals, Top VCs Warn

AI startups are landing pilots at record speed and inflating their revenue numbers with them, three investors and founders warned last week at TechCrunch Disrupt. The panel — Anu Bharadwaj, partner at Iconiq; Rajeev Dham, partner at Sapphire Ventures; and Rahul Vohra, founder of Superhuman Mail — argued that product-market fit has become harder to prove, not easier, in the AI era.

The core problem: pilots are easy, production is not.

Why do pilots distort revenue numbers?

Bharadwaj put the mechanism plainly. "It's never been easier to gain a bunch of pilots because lots of companies across the board are open to trying new things," she told the audience. "But going from pilot to production is really the hard thing."

The gimmick works like this. A founder counts a pilot customer, who may only use the product for a couple of months, inside an annualized run rate figure. That implies the income will continue for a full year. The result looks like a bigger business than the one that exists.

Much of the recent chatter on X has focused on AI founders pulling exactly this move. The trouble starts with the acronym itself. ARR can mean "annualized run rate" or the firmer "annual recurring revenue," and founders often pitch the number without saying which one they mean.

"Is it true ARR or is it run rate?" Bharadwaj asked. "There's many ways to obfuscate the whole thing."

Her advice to founders cuts against the short-term incentive. "It's definitely in your long-term interest that you be very honest and upfront about what quality of revenue this is," she said.

What separates good revenue from hype?

Bharadwaj told founders to separate hype from reality when they examine their own revenue quality. Two patterns raise red flags:

• Only a few customers make up the bulk of revenue.
• Customers can easily switch to another vendor offering a better price.

"Churn rate and retention [are] really dependent on that," she continued.

Vohra, whose Superhuman built a following for its email product, recommends tracking customer engagement and dollar retention — especially for consumer products. "Figure out what benchmark [is] good, great, and excellent for your space," he said.

Superhuman tracked conventional ARR, but it also built an internal metric it called "delighted ARR." Vohra defined it as "the ARR only from customers who are super engaged with the product." The metric showed how sticky customers were and how long they were likely to stay.

"Those users are now still users, those companies are now still customers," he said. "We obsessed over serving that particular category of our user base and our customer base."

When should a founder pivot the whole company?

Vohra offered a numeric test. Talk to customers, he said. Understand what value they find in the product and how they would feel if they could no longer use it. Then measure those responses honestly.

The threshold is unforgiving. "If you're 15%, 14%, 13% or below, consider a complete pivot. Take the codebase, take the team, build something entirely new," Vohra said. "But if you're 20%, you can totally optimize your way from 20% to 40%."

Dham added a warning for ambitious founders: do not confuse having a vision with having a good business.

Do investors expect product-market fit at the seed stage?

No — and that has changed how they evaluate early teams. Bharadwaj said she does not even look for product-market fit at the seed stage. She asks two questions instead: "does the founder have a thesis that seems plausible and do they have the agility to duck and wave as the situation changes?"

Dham goes further. He said he now values founders who have made numerous pivots more highly, because pivots prove the founder has made hard decisions in a market that keeps shifting under them.

The panel's closing message applies to companies at every scale. Even a business with a billion dollars in revenue has not finished the job, Dham argued. "Even at a billion in revenue, these companies are re-earning their product-market fit," he said. "You constantly have to be innovating on products."

For founders raising money in this cycle, the practical takeaway is blunt: disclose which ARR you mean, measure the customers who actually stay, and treat product-market fit as a number you re-earn each quarter rather than a milestone you pass once.