For all their pitches promising something new, AI startups share many of the same questions as startups in years past: How do they know when they’ve achieved the holy grail of product-market fit?
Product-market fit has been studied extensively over the years; entire books have been written about how to master the art. But as with so many things, AI is upending established practices.
“Honestly, it just could not be more different from all the playbooks that we’ve all been taught in tech in the past,” Ann Bordetsky, a partner at New Enterprise Associates, told a standing room-only crowd at TechCrunch Disrupt in San Francisco. “It’s a completely different ball game.”
Top of the list is the pace of change in the AI world. “The technology itself isn’t static,” she said.
Even still, there are ways that founders and operators can evaluate whether they have product-market fit.
One of the best things to watch, Murali Joshi, a partner at Iconiq, told the audience, is “durability of spend.” AI is still early in the adoption curve at many companies, and so much of their spend is focused on experimentation rather than integration.
“Increasingly, we’re seeing people really shift away from just experimental AI budgets to core office of the CXO budgets,” Joshi said. “Digging into that is super critical to ensure that this is a tool, a solution, a platform that’s here to stay, versus something that they’re just testing and trying out.”
Techcrunch event
San Francisco
|
October 13-15, 2026
Joshi also suggested startups consider classic metrics: daily, weekly, and monthly active users. “How frequently are your customers engaging with the tool and the product that they’re paying for?”
Bordetsky agreed, adding that qualitative data can help provide nuance to some of the quantitative metrics which might suggest, but not confirm, whether customers are likely to stick with a product.
“If you talk to customer …