In Plain Sight - 13
In Plain Sight is a weekly blog from Rhys that we will publish every Thursday/Friday. Each edition will be focused on a thorny topic within investing, startups, learning and work and policy.
In Plain Sight is partially a personal attempt to think independently and to avoid leaning too much on AI for answers to our questions. Will we use AI to edit and polish the text? Yes. But more importantly, will we come up with all of the ideas and analysis? Yes.
Here follows Edition 13 - 'Is design converging? Why every car is starting to look the same'
------
Stand on any high street in Europe and try a simple game: cover the badge and guess the car.
A Polestar, a BYD, a Jaecoo, a Volvo and an Audi pull up at the lights. Each is a smooth, high-riding crossover with a thin light bar across the front, flush door handles, a sloping roofline and a paint colour somewhere between white and graphite. Inside, you'll find a large floating screen and very few buttons.
A common reaction from some is a mild lament: car design has lost its soul. Brands have become interchangeable. Nobody takes risks any more (the team behind Jaguar's Type 01 may have something to say about this).
But that reaction may miss something.
It's tempting to blame convergence on lazy designers. In our view, the more interesting explanation is the opposite: products look alike because we've become very good at understanding people.
Over the last few decades, behavioural science has built a growing map of how we actually decide: loss aversion, default effects, anchoring, social proof, variable rewards and the pull of a streak we don't want to break. Kahneman and Tversky gave it academic weight. Thaler and Sunstein's Nudge took it into government. BJ Fogg's lab at Stanford and Nir Eyal's Hooked turned it into a product manual.
Here's the catch. Once a finding is widely known, it stops being an advantage and becomes a best practice. And best practices spread.
If everyone knows a progress bar lifts completion, every onboarding flow gets a progress bar. If everyone knows a red badge pulls us back, every app gets a red badge. If everyone knows we anchor on the middle option, every pricing page shows three tiers with the middle one marked "Most popular".
A/B testing finishes the job. Thousands of teams run thousands of sensible experiments on the same human brain, and they climb toward the same answer.
So behavioural science is not a moat. Over time, it becomes a template. Which raises a question: if our understanding of people keeps improving, will every product eventually look the same?
Let's return to cars...
Look closer and the sameness has several causes, most of them rational.
- Behaviour: Every brand runs clinics on the same buyers and learns the same things: people want to sit high, feel safe and see a big screen. The same insight, applied to the same customer, produces the same crossover.
- Physics: Electric range depends heavily on aerodynamics. Every wind tunnel and simulation pushes toward the same smooth, tapered shape.
- Regulation: Pedestrian-safety rules, crash standards and lighting laws narrow what a front end can look like.
- Shared ownership: Geely owns Volvo Cars and a large stake in Polestar, and shares platforms, parts and engineering across its brands.
None of these is a bad decision on its own. Together, they are a funnel towards a similar end.
Cars are not an usual case. Once you notice it, you see it everywhere. Kyle Chayka called it "AirSpace": the reclaimed-wood, Edison-bulb coffee shop that looks the same in Lisbon, Seoul and Brooklyn. Alex Murrell's essay The Age of Average traced the same pattern across interiors, brand logos, film posters, architecture and, yes, cars.
The pattern is consistent. Markets that get better at measuring what people want tend to converge on it. Optimisation rewards the average, because the average is what the data can see.
If cars took two decades to converge, we suspect software built on AI will take two years.
Consider a typical learning or workforce startup in 2026. It builds on one of a handful of foundation models. It uses the same design systems and components. It borrows engagement mechanics from Duolingo, and its team asks the same AI assistants for positioning, copy and product ideas. Those assistants were trained on the same internet and tend to suggest the same sensible answer.
The result is a new kind of sameness: AI tutors that sound alike, skills platforms with near-identical dashboards and onboarding flows you could swap between products without anyone noticing.
The cost of building a competent product is falling fast. So is the cost of building the same competent product as everyone else.
We find it helpful to think of two paths.
Optimised products start from the data. They ask what the average customer responds to and deliver it well. They are rarely bad, and they're usually fine. They also tend to meet in the middle, and compete on price and distribution once they get there.
Opinionated products start from a point of view. They make choices the data wouldn't have made and accept that some people will dislike them. Think of Tesla's Cybertruck, the transparent design of Nothing's phones or Duolingo's owl. Each of these was a bet, not a finding.
The irony is that the better behavioural science gets, the more valuable the second path becomes. When every product converges on the optimum, the scarce thing is no longer knowing what customers want. It's having the conviction to give them something they didn't know to ask for.
Sameness is where a market settles. It's also the opening for whoever breaks from it.
- Convergence is often good for customers. Similar cars are safer, more efficient and cheaper to build. Familiar app patterns are easier to use. Not all sameness needs fixing.
- Being different is not the same as being better. For every Cybertruck there are many bold products that simply failed.
- The playbook wears out. When every app sends streak reminders, people tune them out. Notification fatigue suggests the known tricks lose power as they spread, which may push teams back toward novelty.
- Divergence may just move. Maybe products converge while brand, community, service and price become the real points of difference.
The last point may be the most important. The question may not be whether products converge, but where differentiation goes once they do.
The cars on the high street will probably keep looking like Polestars for a while. The companies we're most excited about are the ones that don't...
If you're building an opinionated product in learning, work or AI, we'd love to talk to you.
