The most dangerous startup metric is often the one that looks cleanest on a slide. Annual recurring revenue feels like proof because it has a dollar sign, a growth curve, and the comforting smell of inevitability. But in the current AI cycle, where a clever workflow can be cloned before the sales call follow up hits the inbox, early ARR can be less a scoreboard than a fog machine.
The 90 day scoreboard is changing
TechCrunch reported that at TechCrunch Disrupt, Lovable go to market lead Luke Arno, Polsia founder Ben Broca, and Theory Ventures investor Tomasz Tunguz discussed what founders should do in the first 90 days after a commercial product is ready to sell. That framing matters because it shifts the launch question from, who paid, to, what did we actually learn that can be repeated? It is the difference between finding a market and collecting interesting anecdotes with invoices attached.
TechCrunch also reported that Broca described Polsia as an AI platform that uses agents to build and run businesses with almost no human staff, and said he is the company’s only employee. His point was not just operational minimalism. It was that product market fit is harder to pin down when many products can be copied by competitors, which turns early traction into a weaker signal than founders may want it to be.
@title Three lenses on the 90 days
@source TechCrunch
Person Role Signal
Luke Arno GTM lead repeatable wins
Ben Broca Polsia founder copy risk
Tomasz Tunguz Theory Ventures capital pressure
@caption The panel framed early growth around repeatability, copy risk, and execution pressure.
That is the competitive landscape map hiding inside the panel: AI lowers the cost of building, which also lowers the cost of imitation. If the product can be copied, the moat has to show up somewhere else, often in distribution, workflow ownership, data feedback, or a use case wedge that customers keep pulling into the company. A first customer is evidence. A repeated path to the next customer is a machine beginning to hum.
Early ARR can be a sugar high
TechCrunch said Lovable is valued at more than $13 billion, says its ARR has passed $600 million, and has raised more than $700 million. Those are extraordinary numbers, and Tunguz called Lovable’s growth exceptional while also noting that execution pressure is very high. TechCrunch also reported his view that capital is far more available now than when he began investing in 2008.
That combination is the AI startup version of being handed the keys to a race car before proving you can take a corner. Capital lets a company hire, expand, and sell faster, but it does not automatically tell the team which motion is real. If anything, abundant capital can make the signal noisier because founders can buy activity before they understand repeatability.
Arno’s warning, as reported by TechCrunch, is the useful counterweight: fast revenue in the first 90 days should not be treated as proof that the company is ready to scale. That is not anti sales. It is pro diagnosis. Early ARR is a receipt; product market fit is a repeatable explanation for why the receipt happened.
Learning density beats sales theater
TechCrunch reported Arno’s advice that founders should build “learning density” before rushing into a top down go to market motion. In plain PM terms, this means every customer conversation should shrink uncertainty, not just add another logo to the pitch deck. A founder should know which buyer felt the pain, which use case converted, what language resonated, and which part of the sales path can survive contact with customer number four.
Arno also told founders that durable go to market strategies are ones that can be “reliably repeated,” according to TechCrunch. His cleanest diagnostic was the contrast between landing 10 customers in 10 different ways for 10 different problems and landing three customers through a similar motion. The former is revenue. The latter is the beginning of a sales system.
This is where founders should be especially careful with AI positioning. If the pitch is broad enough to sell to anyone, it may be too broad to teach the company anything. A good first 90 days should feel less like a victory lap and more like sorting receipts after a messy trip, tedious, revealing, and absolutely necessary before approving the bigger budget.
The next logical move
Read through TechCrunch’s account and the incentive structure becomes pretty clear: the next smart move for AI founders is not to hire a large sales team the moment ARR twitches upward. It is to instrument the launch like a learning loop. Track where customers come from, what problem they share, what objection repeats, which demo moment lands, and whether the same path works twice.
For investors, the diligence question should evolve too. Do not only ask how much ARR is in the spreadsheet. Ask how much of it came from the same motion, the same buyer, and the same problem. If every deal needed a custom story, the company may have a consulting pattern dressed up as software.
For builders, the practical takeaway is simple: spend the first 90 days proving the motion, not celebrating the metric. Early sales still matter, but their job is to teach. The founders who win this cycle will be the ones who turn those first messy wins into a repeatable path before competitors turn their product into a feature.