The opportunity after software.
As intelligence becomes abundant and cheap, the next real opportunities move toward workflows, trust, and decisions.
When intelligence becomes cheap and widely accessible, simply adding AI to a product stops being a differentiator — the opportunity moves to workflow design, trust, and context. Software made information programmable. AI is making parts of intelligence itself more accessible, and that’s a genuinely different kind of shift than any single feature or product category.
Where value moves once intelligence is abundant
That changes where value gets created. If intelligence becomes cheaper and easier to access for everyone, adding intelligence to a product isn’t enough on its own anymore — everyone has access to roughly the same underlying capability, which means the capability itself stops being the thing that differentiates one product from another.
What starts to matter more, once the raw capability is commoditized, is everything around it: how well a product understands the specific context of a specific user’s situation, how much it can be trusted with a consequential decision, and how well it fits into a workflow the user already has rather than demanding a new one.
The real question worth asking
The real question becomes: what should happen differently now that this capability exists? Not “can we add AI,” but “what changes because we can.” That’s a much harder question to answer well, because it requires genuinely rethinking a workflow rather than bolting a new capability onto an old one.
Most early attempts at this skip the hard version of the question and answer the easy one instead — they keep the existing workflow exactly as it was and insert an AI-powered step somewhere inside it. That can produce real value, but it rarely produces the full value available, because the old workflow was often built around the previous set of constraints, some of which no longer apply.
What this could reshape
This could reshape how people create, search, plan, learn, run businesses, and make decisions day to day — often in ways that don’t look like “AI features” at all. The most significant changes tend to be the ones that don’t announce themselves as AI-related, because they’ve been absorbed so completely into how a task gets done that pointing at the AI part feels almost beside the point.
That’s worth sitting with for anyone building product right now: the most defensible position isn’t “we have AI.” It’s “we understood this workflow well enough to redesign it properly once AI made the redesign possible.”
Common questions
If AI capability becomes commoditized, how does a product stay differentiated? Through workflow design, trust, and context — the parts of the experience that require understanding a specific user’s situation deeply, which a general-purpose capability alone doesn’t provide.
Does this mean the underlying AI model matters less over time? For most products, yes, relative to how much it seemed to matter early on. The model becomes closer to infrastructure — necessary, but not sufficient to explain why one product wins over another doing something similar.
What’s the risk of redesigning a workflow too aggressively around new AI capability? Losing users who relied on the old workflow’s familiarity. The right pace usually involves proving the new approach alongside the old one before fully replacing it.
Takeaway: we’re interested in this transition because it creates a new generation of product questions — not just technical ones, but genuinely human ones.