In recent years we’ve confused product-market fit with virality — especially in the AI world, where a product that racks up millions of views immediately spawns the ‘game changer’ narrative, only for six months to pass and nobody to be using it anymore.
We’ve forgotten that real product-market fit isn’t measured in attention but in staying power. That’s why I believe OpenAI and Anthropic may have found their product-market fit in AI development agents: they’re maturing, adoption is broad, there’s still meaningful room for improvement, and the token-driven revenue they generate means Anthropic could soon post its first profitable quarter, with OpenAI close behind.
People try an enormous number of digital tools but adopt very few — and when they do adopt one, it works its way into their daily routine. That’s exactly what’s happening with development agents, which, unlike simple chat interfaces, have the potential to generate serious revenue.
Attention and value are two completely different economies. My strong suspicion is that instead of creating value, many companies today are competing for attention by slapping the word AI onto their marketing.
A word of warning: AI-washing doesn’t produce the kind of positive dependency that builds successful software products. That’s why I think the coming years will bring a sharp divide between AI-enabled products and AI-infrastructural products.
The former will keep generating hype and squeezing residual value out of their market slice — many will probably fail. The latter, by contrast, will be absorbed into daily work. Just as ’to google’ became a synonym for ’to search,’ AI-infrastructural products will become synonymous with the actions they perform and embed themselves in pop culture.