One person, a whole team
A product's full chain, from paper to production, now fits in one person's hands. That changes team design.
If you never sat through the start of a digital product, you may not know what it used to cost. Development environment, repository, infrastructure, deploy pipeline: before any user saw any screen, weeks of specialized work. DevOps for infrastructure. Backend for the foundation. Frontend for the interface. A designer for the flow. A manager to coordinate everyone. A product's initial setup demanded a whole team, and the whole team was expensive before there was any evidence the product deserved to exist.
Today one person does it. With AI, one person covers the full chain: forms the hypothesis, designs the user journey, defines the MVP, develops with Claude Code or similar, deploys to the cloud, instruments the data with GA4, GTM and BigQuery, and handles SEO and GEO so the product gets found by people and by machines. From paper to production. And I don't mean a demo prototype: I mean a product in production, with real users, telemetry on and a deploy routine. I know because that's how I work: every product I ship goes through that entire chain, in my hands.
This changes team design, but not the way the hype suggests. The product, design and engineering triad still makes sense. What changes is the condition: it holds from a certain scale and complexity up, not as a universal default. For a simple product, one builder is enough: a person with product judgment, from any background, who has mastered the tools. That could be a PM who learned to build, a designer who learned to structure data, an engineer who learned to talk to users. For a complex product, specialists are still needed. But even there the work changes: far less backlog, far less coordination bureaucracy, because each specialist carries more of the chain alone. The right question is no longer "how many people does this product need" but "at what point does it need more than one".
What about process? That's the part that interests me most. In place of heavy process, non-negotiable values defined per product. Mine: every change documented, no exceptions. Security by design, not as an audit afterwards. User data untouchable, never used beyond what was agreed. The AI squad, me and the tools, is steered by those values, not by ceremonies. I don't need a daily standup to align with myself. I need clear principles so that speed doesn't turn into carelessness. Those values do the job process used to do, creating predictability, at close to zero overhead.
One honest note before closing: this is new. Very new. Everyone is still learning what works in this model, myself included. Some weeks I discover that a flow I considered solved needs redoing. Some tools promise and don't deliver. What I write here is a snapshot of what I see working now, not finished doctrine. But the direction seems clear to me: coordination cost was product development's biggest hidden tax, and it is collapsing. Whoever learns to work this way first gets ahead.