AI is putting software in more hands than ever. When every team can build its own solution, someone still has to make the product work as a whole.
A product manager can now prompt a working interface into existence. In a randomized trial Figma ran on its own prompt-to-design tool, 50 product designers and 50 product managers completed standard design tasks about 20% faster with the tool, and the product managers gained more than the designers did (arXiv). At Coinbase, agents now read customer feedback, prioritize bugs and features, draft the code and hand it to an engineer for final review, every day (Brian Armstrong). Figma's chief design officer told TechCrunch that 17 times more lines of code are being produced than in the past, but only 1.3 times more releases (TechCrunch).
Those numbers measure different things. Together they describe a shift in who builds: more people, more roles and more automated loops can now produce working pieces of a product. None of them describe who is responsible for how those pieces add up.
That's the gap. Building a feature and owning a product were never the same job. For much of the last fifteen years, building was slow and expensive, so changes passed through a small number of people and processes. That didn't guarantee a product that held together, but it limited how fast disconnected ideas became working software. Building is getting cheaper fast. Responsibility for the whole hasn't moved with it.
We've been here before
This isn't the first time the ability to build outran the ownership of what got built. Jakob Nielsen estimates the UX profession grew from about 1,000 people in 1983 to about a million by 2017 (Nielsen Norman Group). His explanation of why matters more than the numbers. Before personal computers, the people who bought software often weren't the people who used it, so how it felt to use rarely decided the sale. Personal computers changed that: the person choosing the software was now the person using it. The web went further. Before anyone could pay a company online, they had to find their way through its website, and if they couldn't, the sale never happened. Once the experience helped decide whether a customer paid, companies had a business reason to take it seriously.
I started in creative direction in 1999 and moved into product design in 2013, so I watched the back half of that shift from inside it. Much of the software of that era was shaped by the people who built it, around technical function and business process. In my experience, what changed first wasn't how many designers a company hired. It was that someone in the organization became answerable for how the product worked for the people using it, from start to finish. Design teams, research practices and new job titles grew up around that responsibility.
More designers didn't guarantee better products, either. Plenty of companies hired a design team and still shipped products nobody owned from end to end. The products that improved were the ones where a specific person or people were responsible for the whole experience.
What it looks like when nobody owns the whole
When I became Head of Design at Flowbird in 2022, its Hub dashboard, which city operators use to run parking across their networks, surfaced data those operators couldn't use effectively. Design leadership hadn't been involved in its interface. The design team worked as a production service: product owners requested screens as they needed them, often driven by sales, with no research, no backlog and no review of what reached development. Every screen made sense to whoever asked for it. Nobody was responsible for how they added up.
The first fix wasn't a redesign. It was a change in who answered for what. Research and testing became part of how products were made, not an optional step. Product managers joined every planning ceremony and shared ownership of outcomes with design. Engineers saw work in design reviews before it was built. Loose month-long cycles, agreed informally between individual designers and product managers, became two-week sprints with defined milestones. Handoff to engineering, which had been a black box, got a set process and a follow-up review before anything reached the market.
Over the next two and a half years, the design organization grew to ten people across several countries. Average design cycle time fell from about four weeks to 2.2, and team retention rose from about half to 95%, a level it held through the company's merger with EasyPark. Those results came from several changes working together, not one. All of them depended on someone being answerable for how the work fit together.
Estate Guru showed the same pattern in a regulated setting. The platform serves advisors, attorneys and the families they work with. When I started, there was no product design team and one junior product manager, and the experience had been shaped almost entirely by engineering, for sound engineering reasons. Research showed what that cost. In six of ten interviews, advisors couldn't tell when documents had shipped, and more than 60% of participants couldn't find key functions within three clicks. In a product where a missed step can be expensive and hard to undo, these aren't just usability problems. They're symptoms of an ownership problem: nobody was accountable for whether the experience worked from start to finish. Once someone owned the whole experience, task completion improved 35% and errors fell by over half.
What owning the experience means now
Ownership isn't approval rights. It's a line of accountability that runs from what the product is for to what actually ships, and it has three parts.
Start with what the product is for. Who it serves, what it has to accomplish, and in what order. I begin every engagement there, from the market inward rather than the feature outward. When anyone can build, a written statement of intent is what lets them build in the same direction.
Make it add up. Someone has to be responsible for how features, workflows and interfaces fit together, including work that starts outside design: a product manager's prototype, an engineer's fix, an agent's pull request. That person doesn't have to make every piece. They have to know how every piece lands.
Check it against reality before it ships. Review should be proportionate to risk, and evidence should decide what moves forward. At Flowbird, when we brought AI into research synthesis, it cut interview analysis from two or three days to four to six hours. We kept one rule: no AI-generated insight informed a decision until a person had verified it.
Where this doesn't apply
Not every product needs a formal owner. Prototypes, internal tools and narrow single-purpose products can live without one.
Some teams already hold the whole without a gatekeeper. At Anthropic, every designer pushes code in their first week (Designer Fund). At Linear, the CEO built an AI skill trained on the company's own writing about how it thinks, so the tools people use carry the company's intent with them (Designer Fund). Ownership there lives in people and practice. That's the point: ownership doesn't require a title or a sign-off step. It requires a responsibility someone actually exercises, whether that's a founder, a product manager, an engineer or a designer.
This isn't a prediction about design hiring. The argument holds whether design teams grow, shrink or stay the same size. And the velocity figures above are early. Figma's trial measured structured tasks on its own tool, and nobody has yet measured what all this new output does to products over time.
The part nobody was assigned
Building will keep getting easier. Owning the experience will stay a choice. When building is cheap, the expensive part is the one nobody was assigned.
References
Stewart et al., “Does AI Save Time on Product Design? A Randomized Controlled Experiment of AI Prompt-to-Design Workflows,” arXiv, September 2026.
Brian Armstrong, post on Coinbase's daily agent loop, X, July 14, 2026.
TechCrunch, “Figma announces new features to define how agents work on its platform,” October 14, 2026.
Jakob Nielsen, “A 100-Year View of User Experience,” Nielsen Norman Group, 2017.
Designer Fund, “How tech's most craft-obsessed team uses AI: Inside design at Linear,” August 12, 2026.
Designer Fund, “When code is no longer the constraint: Inside Anthropic's design team,” October 6, 2026.
