Design From the Inside (If You're Allowed To)
Matt Ström-Awn’s tactics are correct. The unstated precondition is who gets to use them.
Matt Ström-Awn published a piece arguing that product designers in the AI era should stop thinking like architects. The architect designs from outside the building, using master plans and elevations to coordinate engineers around an ideal reality. That model breaks when engineers can build so fast and so independently that no master plan stays current for more than an afternoon. Ström-Awn’s alternative is a designer who works inside the building, laying high-visibility tape on the floor to nudge traffic, knocking down walls when nobody’s watching, behaving as if the process is already in place rather than announcing one and waiting for adoption.
The piece is correct about the tactics. I want to add the psychological layer Ström-Awn implies but doesn’t name, because that layer is what determines whether his argument is a craft prescription or an unintentional filter for who is allowed to do good design work.
The tactics are well-grounded behavioral psychology
Most of what Ström-Awn describes maps onto a literature he isn’t citing.
The floor tape is choice architecture. Richard Thaler and Cass Sunstein’s Nudge is the canonical text on the principle Ström-Awn is enacting: small changes to the environment produce larger behavioral shifts than direct persuasion does. The “wet floor” sign in front of the redundant bathroom is a textbook nudge. So is widening the hallway by a few inches at a time, “staying just on the edge of notice.” These work because they bypass the deliberative system and operate on the automatic one. The staff doesn’t decide to walk a more efficient path. They just walk it.
The "ritual, not process" point connects to a deeper finding. Wendy Wood spent three decades studying why environmental design beats announced policy, and the core argument in Good Habits, Bad Habits is that habits form through repeated behavior in consistent contexts, not through intention or deliberation. When the context is right, the behavior becomes automatic. When it isn't, even strongly held intentions don't translate. Sendhil Mullainathan and Eldar Shafir's work on Scarcity adds the organizational layer: when people are operating under cognitive load, which is most people, most of the time, in most growing companies, deliberation is the first thing to go. New processes require deliberation to adopt. Rituals become the path of least resistance. "Behave as if the process is already in place" works because it removes the deliberation step entirely. The team doesn't have to decide to use the PRD format. The PRD is just there.
The desire paths argument ("manifesting desire paths into the world") has a well-documented empirical basis, and it isn't really Hillier's. The concept comes from landscape architecture and urban planning, where the gap between intended routes and actual routes has been studied for decades. The Wikipedia entry on desire paths captures the core finding cleanly: when a path diverges from the design, it's usually treated as user error, when design theory suggests it should be read as feedback. The canonical example is Ohio State University's Oval, where campus planners deliberately withheld paving until students had worn their preferred routes into the grass, then paved what was already there. Michigan State did the same thing in 1914: the university architect waited for winter, went up in a hot air balloon, mapped the trails students had cut through the snow, and used those as the basis for the permanent walkway system. Ström-Awn's instinct to read traffic patterns over interview data is the same move, transposed to product design.
So the tactics work, and they work for reasons the cognitive science literature understands.
What he doesn’t say about who gets to do this
Every tactic in the piece presupposes a particular structural position. The designer who buys the safety tape and lays it on the floor has, implicitly, the authority to redirect traffic without permission. The designer who knocks down walls “completely unnoticed” is operating in a context where unilateral action is either tolerated or actively rewarded. The designer who behaves “as if the process is already in place” is making a bet that nobody will challenge their authority to install a process the company never agreed to.
Those bets are not equally safe for everyone making them.
There's a body of organizational psychology research on what's called proactive behavior, originating with Morrison and Phelps' 1999 paper in the Academy of Management Journal, which defines it as voluntary, constructive efforts to initiate change beyond one's formal role. The follow-up literature on how that behavior gets evaluated is where the relevant finding sits. Clarissa Bohlmann and Hannes Zacher's 2021 study in the Journal of Business and Psychology used experimental vignettes to test how observers rate identical proactive behavior depending on the actor's age and gender. Across two studies with more than 260 raters generating nearly 3,000 individual ratings, achievement-motivated proactive behavior was rated as more effective for older men than for younger men. Younger women received higher effectiveness ratings than older women regardless of motive. Senior people performing unilateral action get credited for showing leadership. Junior people performing the same action get coded as overstepping. People from groups whose authority to act independently isn't culturally expected pay an additional cost on top of that.
I wrote about a related dynamic in the likability tax. The same boundary-shaped behavior reads as decisive from one leader and obstructive from another, and the difference turns on gender, seniority, or fit with the dominant cultural pattern of the room. Ström-Awn’s “design from the inside” is structurally similar: the same tactics that get a senior designer at a Series B startup praised as visionary will get a junior designer at the same company labeled as “not staying in their lane.”
This is not a hypothetical concern. Recent research by David Zweig and colleagues at the University of Toronto has established knowledge theft as a measurable, distinct form of workplace deviance. Across studies with more than 1,500 workers in the US, UK, and Canada, 91% of participants reported being a victim, perpetrator, or witness of having their ideas or work taken. The follow-up study traces the emotional consequences and shows that misattribution operates differently when the misattributor holds higher status than the original idea’s source. When a high-status person ships a redesign without permission and it works, the company learns a positive lesson about boldness. When a lower-status person does the same thing, retroactive credit reassignment is one of the documented patterns: the redesign gets remembered as someone else’s idea, and the original designer disappears from the credit chain.
So when Ström-Awn writes “don’t ask for permission or seek consensus; design the product from the inside,” he’s giving good craft advice that carries, for a meaningful subset of the people reading him, an unstated precondition: if your structural position permits it.
The shadow side of ritual-not-process
Ström-Awn’s “ritual, not process” insight is correct, and it’s also the move with the highest variance. When it works, it works beautifully. When it fails, it fails in a particular shape the piece doesn’t address.
A designer behaves as if the process is in place, the company tolerates it for a while, and then something breaks. A redesign causes a regression. A unilateral PRD format upsets a stakeholder who wasn’t consulted. A floor tape redirect routes traffic in a way that breaks a workflow nobody told the designer about. At that point, the question becomes: was this initiative or was this overreach?
The answer is almost never determined by the merits of the action. It’s determined by the social capital of the actor. A senior designer with a track record of wins gets credited with bold experimentation that didn’t pan out this time. A junior designer, or a designer from an underrepresented group, gets coded as having confirmed concerns that were already there.
The fundamental attribution error is the mechanism: named by Lee Ross in 1977, it describes our tendency to over-attribute other people's behavior to their character and under-attribute it to their situation — while for our own behavior, we do the reverse, reaching for situational explanations first. When the dominant-status designer’s experiment fails, observers default to situational attribution (”the timing was off”). When the lower-status designer’s experiment fails, observers default to character attribution. The mechanism is automatic and largely invisible to the observers themselves.
This is why the same advice works for one designer and damages another.
What the less-protected designer can do
The honest answer is that “design from the inside” is excellent counsel for designers operating in environments where their authority to take initiative is structurally protected, and considerably riskier for designers operating in environments where it isn’t.
A few things that help close that gap, in order of how much they require from the organization versus the individual:
Build documentation as a defensive measure before the tape goes down. A short Slack message to a manager (”I’m going to try X for the next two weeks based on Y observation, will report back”) doesn’t compromise the inside-out approach. It does protect against the attribution asymmetry that fires when something goes wrong. The senior designer relies on social capital to frame unilateral action as initiative. The less protected designer needs a paper trail that does the same framing work explicitly.
Use the ritual move alongside someone with cover. Ström-Awn’s “behave as if the process is in place” works much better when a senior person is visibly doing the same thing. The ritual becomes legible as a team norm rather than one person’s idiosyncratic behavior. The McKinsey Women in the Workplace 2025 report puts numbers on this: at entry level, 45% of men report having a sponsor compared to 31% of women, and the gap compounds going up. Designers who are given sponsorship naturally can act without thinking about it. Designers who aren’t have to construct it deliberately.. Designers who are given it naturally can act without thinking about it. Designers who aren’t have to construct it deliberately.
Know your reversibility threshold. Inside-out tactics work best for decisions that can be quietly undone if they don’t pan out. Floor tape can be peeled up. Ström-Awn treats all his examples as low-cost, but for designers without protective social capital, the threshold for which decisions are safely “inside-out” sits lower. Knowing where your threshold is matters more than knowing where it sits in the abstract.
And track what gets credited. The most insidious failure mode isn’t visible action being punished. It’s invisible action being uncredited. The designer who quietly ships improvements that work, and watches a senior colleague get credit for the resulting metric movements, is paying a cost the piece doesn’t address. Timestamps in a shared doc aren’t bureaucracy. They’re an audit trail that makes invisible work countable.
What I’d add
Ström-Awn is making a craft argument in a craft register, and the craft argument holds. What I’d add for the engineering and design leaders in my audience is this: the tactics in this piece operate in a particular interpretation environment, and that environment isn’t neutral. If you’re a senior designer with social capital, use them as written. If you’re managing designers, your job is to build the conditions in which the rest of your team can use them safely too, which means defending unilateral action when it produces good outcomes regardless of who took it, attributing wins accurately when the inside-out approach pays off, and recognizing that “she should have asked first” is the failure mode where the advice you gave her gets retroactively converted into evidence against her.
The architect-versus-inside-out frame captures something real about how design has to work when the building keeps changing while you’re in it. The question it leaves open is who gets to be the inside-out designer, and what it takes to make that role available to more people than it currently is.
The tactics are available to anyone. The permission to use them isn’t.