A few months ago, we talked about how AI would change the way software gets built, and at the time it still felt like something slightly abstract, interesting to think about but not yet forcing immediate decisions. That has changed. AI-native companies are now shipping faster, with fewer people, and at a fundamentally different cost structure, and they are no longer operating in a separate experimental space but competing head-on with established SaaS companies. The question is no longer whether this matters, but whether existing organizations can adapt fast enough. What AI-native actually means AI-native does not mean that a company has added AI features to an existing product or sprinkled AI into a few workflows. It means the company could not have been founded before this wave of AI, because the way work gets done would simply not have been possible. In these companies, the first line of code is written by an agent, documentation and contracts start with AI drafts, research and support replies are generated by default, and humans mainly step in to review, correct, and provide direction. Using AI is not a strategic decision anymore; avoiding it requires justification. Why this is hard for legacy companies Most legacy software companies are not badly run, and the structures they operate with were rational responses to the constraints of the past. They hired large engineering teams because building software was slow and expensive, they introduced product managers to translate customer needs into specifications, and they added layers of coordination because that was the only way to scale delivery reliably. AI changes one thing fundamentally: the cost and speed of translating customer intent into working software. What used to require weeks of alignment, handoffs, and planning now often happens within hours, but organizations are still optimized for the old reality. The people problem AI-native companies hire very differently from traditional SaaS organizations. They delay hiring for as long as possible, prioritize people who can both think and build, and try to avoid roles whose primary purpose is coordination rather than value creation. Legacy companies already have many of those roles in place. As a result, pushing AI adoption often means questioning why certain jobs exist at all, which is uncomfortable, politically difficult, and often avoided, leading transformation efforts to slow down or stall entirely. Distribution isn’t enough anymore Established companies often argue that their main advantage lies in distribution, and historically that has been true. However, AI-native companies operate with a cost structure that allows them to price their products very differently, because a team of ten can now compete with what previously required a team of one hundred. That pricing flexibility itself becomes a powerful distribution mechanism, and it is extremely difficult to out-market a product that is cheaper, improves faster, and is built by a much leaner organization. Product management changes the most The role that changes most visibly in AI-native companies is product management. Product managers are no longer primarily facilitators or coordinators, but builders who combine customer understanding with the ability to turn insights directly into working software. They talk to customers in the morning, prototype during the day, and often ship something meaningful by the evening, which causes discovery and delivery to collapse into a single continuous motion. This strongly favors people who deeply understand customers and can act immediately, while it puts pressure on roles that exist mainly to coordinate work between others. Fewer people, more context As execution becomes cheaper, context becomes more valuable. The most important assets of future software companies are the codebase itself, rich and ongoing customer conversations, and clearly written strategy, constraints, and decision principles that guide both humans and AI systems. AI systems need this context to act well, and humans do too, which creates a new leadership responsibility focused less on managing output and more on maintaining shared clarity. Can legacy companies win? In theory, legacy companies have everything they need to compete. They have distribution, long-standing customer relationships, and people with deep domain knowledge who understand the problems better than any newcomer. If they manage to radically reduce coordination overhead, turn product managers and engineers into customer-centric builders, and accept smaller, more empowered teams, they can remain competitive. In practice, this is rare, not because leaders are incompetent, but because the organizations they built were optimized for constraints that no longer exist. What stays true Despite all the change, some fundamentals remain. Customer understanding still compounds over time, clear strategy still matters, and good judgment is still scarce. What disappears is the need to scale through headcount. Software companies will likely become smaller again, not simpler, but leaner, and very different from what most of us are used to. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit thehotfixpodcast.substack.com
