It’s an Input, Not an Instruction
What a bird in the hand taught me about territorial design thinking
Most designers have been sent a layout they didn’t ask for.
Maybe a PM sketched the flow, a founder assembled a rough homepage, or someone used AI to show you what they were imagining instead of explaining it. I’ve received plenty of these over the years, and I’ll admit my first reaction hasn’t always been gratitude. There’s a small defensive flicker that can happen when someone wanders into your discipline:
Eventually I realized that reaction was confusing two things. They hadn’t finished my work; they’d given me a starting point.
A bird in the hand
“A bird in the hand is worth two in the bush” is usually advice about risk: take the sure thing over the possibility of something better.
I’ve started thinking about it as a decent design principle, too. A rough artifact that exists can be more valuable than a perfect one that doesn’t.
Someone hands you a wireframe, a PM sketches a flow on a whiteboard, or a CEO generates a questionable logo because pointing at something is easier than describing what they’re imagining. None of these need to be shippable, or even good. They need to make thinking visible.
Before the artifact exists, someone might tell me they want something “premium,” “editorial,” “approachable,” “clean,” or any of the other adjectives designers spend their careers translating into decisions. Then they show me something, and suddenly I can see what premium means to them. I can see the hierarchy they’re imagining, the density, the tone, what they’re emphasizing and what they’re ignoring.
I might disagree with every decision they made, but the ambiguity collapsed. A rough artifact makes someone’s assumptions visible, whether or not it’s right. Once those assumptions are visible, I can interrogate them: preserve what works, challenge what doesn’t and understand the intent underneath both.
That’s more useful than designing against adjectives.
Nobody owns the starting point
I used to think expertise meant owning the first move. If it’s a design problem, the designer generates the first draft; if it’s copy, the writer writes it. Anything else can feel like someone stepping on the discipline.
I’ve come to think that’s backwards.
Expertise is what you do with a rough input, whoever produced it.
A PM who sketches a flow is externalizing an idea. A founder who sends over an AI-generated homepage might simply be saying, this is closer to what I mean than anything I’ve explained out loud.
It’s an input, not an instruction.
I don’t have to preserve it, like it or use any of it, but I know something now that I didn’t five minutes ago. That’s the bird in the hand. The two in the bush are the theoretically superior artifact that might have existed if everyone stayed inside their job description and waited for the designated expert to make the first move.
Iteration is the profession
Designers rarely begin with nothing.
We inherit existing products, brand systems, research, analytics, customer complaints, technical constraints, competitive patterns, stakeholder ideas and other people’s half-finished sketches. We inherit decisions we agree with and plenty we don’t. Then we make the next version better.
Early on, it’s tempting to measure your contribution by how much of the artifact you personally created. Over time, I’ve found a better measure: how much did the outcome improve because you were involved?
Those aren’t the same thing.
If someone else gets us 20% of the way there, my value isn’t diminished because I didn’t produce the first 20%. My responsibility is figuring out what the remaining 80% requires. Sometimes that means refining the idea, sometimes dismantling it, and sometimes discovering we were solving the wrong problem entirely.
Great design has never required starting from zero. One of the more valuable forms of expertise is knowing what to do when the canvas is already messy.
The same rule has to work both ways
The contradiction becomes obvious when I consider how often I wander into adjacent disciplines myself.
Designers do this constantly. We comment on copy, question strategy, suggest product decisions, challenge technical assumptions and occasionally make something that technically belongs to someone else’s discipline. I’ve done it too.
AI has made that kind of exploration easier. I can take an incomplete thought further before handing it to someone who knows more than I do. I can test an argument, explore language, interrogate an assumption or turn something half-formed into an artifact another person can respond to.
That doesn’t make me a copywriter, strategist or engineer. It doesn’t need to. But it does force me to reconsider territoriality around my own discipline. If I value the freedom to explore adjacent fields, I have to extend the same generosity when someone explores mine.
Crossing a boundary doesn’t erase it. The specialist stays valuable because they can take the rough thing somewhere the generalist can’t.
AI changed the cost of the first draft
People have wandered across disciplinary boundaries for as long as those disciplines have existed. Stakeholders were sketching interfaces long before generative AI, designers were rewriting headlines, engineers were suggesting interaction patterns, and writers were drawing diagrams. Founders have always made terrible logos.
Turning those thoughts into artifacts got cheaper.
Someone who couldn’t build a convincing interface can now generate something interface-shaped in minutes, and someone uncomfortable with writing can turn an incomplete thought into a rough narrative. That can feel threatening if we confuse producing the artifact with possessing the expertise. They’re not the same thing.
AI lowered the cost of producing something for expertise to react to; it didn’t eliminate the need for expertise itself. The first draft got cheaper. Judgment, taste and context didn’t. Neither did understanding users, systems, constraints and consequences, or knowing which 80% of a generated artifact should be thrown away.
If anything, cheaper production makes those abilities more important. We’re going to have far more material to evaluate, and generating options is not the same skill as knowing which option deserves to survive.
AI is making authorship cheaper while making judgment more valuable.
For designers, that’s a useful distinction. Our value has never been about the speed of putting the first plausible arrangement of rectangles on a screen. It’s about what happens downstream: interpreting, selecting, challenging, connecting, refining and knowing why one decision should survive over another.
Additive or substitutive?
There is, however, a line.
If someone spends hours working outside their discipline and their actual responsibilities suffer, that’s a problem. If a stakeholder creates a mockup and expects a designer to execute it pixel for pixel, that’s not exploration. If someone generates an artifact and treats its existence as proof that the professional work is finished, that replaces expertise instead of augmenting it.
Those situations deserve pushback. They’re different from someone contributing additional thinking.
I’ve found a simpler distinction useful: is the work additive or substitutive?
Additive work says, here’s something that might help.
Substitutive work says, I did your part for you.
A CEO generating an interface to communicate an idea can be additive. Generating one and telling the designer to reproduce it is substitutive. A designer suggesting copy because an interaction isn’t clear can be additive. Rewriting everything, bypassing the writer and declaring the work finished is substitutive.
The artifact doesn’t decide which one happened. The people involved do.
So the useful question isn’t whether someone crossed a disciplinary boundary. It’s what happened when they crossed it.
The Additive Work Test
I’ve started using three questions to make that distinction:
- Did their actual responsibilities suffer?
- Did the extra work create useful information, clarity or momentum that didn’t exist before?
- Did they treat what they made as an input to be challenged and improved rather than a finished mandate?
If the answers are no, yes and yes, there probably isn’t a problem to solve. There’s an engaged colleague contributing beyond the minimum boundary of their role.
Whether they used Figma, a whiteboard, a napkin, ChatGPT or some tool we’ll all be complaining about six months from now isn’t the important part. The useful behaviour is the initiative.
Curiosity is allowed to produce bad work
Creative work has a peculiar professional purity problem. We sometimes treat contribution as more legitimate when it stays neatly inside the contributor’s title, tools and established expertise.
I don’t find that useful.
I’d rather collaborate with the designer who occasionally writes, the writer who sketches, the engineer who cares about interaction design and the PM who opens Figma when explaining something would take longer.
They won’t always be good at it. That’s almost beside the point.
Curiosity is allowed to produce bad work. Expertise gets to make it better.
This isn’t an argument for hustle culture, either. Doing your job well is enough. Nobody owes their employer a portfolio of extracurricular intellectual labour, and working beyond your discipline shouldn’t become an expectation disguised as “initiative.”
But there’s a meaningful difference between not requiring more and discouraging more when someone voluntarily contributes it.
Optional contribution shouldn’t become a liability. If someone’s curiosity produces something useful without compromising what they’re responsible for, I’m not interested in policing whether they were technically the right person to make it.
Don’t optimize for less initiative
The careers page might list values like initiative and ownership, but people learn the real answer by watching what happens when someone ventures beyond the minimum definition of their role.
If the response to useful extra work is territoriality, people learn quickly. They stop volunteering the thought, making the rough draft or exploring the adjacent problem. They wait for the meeting, wait for the owner, and do exactly what was assigned.
Eventually, an organization can become very efficient at producing exactly the amount of thinking contained in everyone’s job description.
That’s probably not something worth optimizing for.
Good collaboration is messy: disciplines overlap and ideas arrive from the wrong people. Designers write, writers sketch, PMs prototype, engineers suggest interaction patterns, and founders make terrible logos. Some of it will be bad, and that’s fine.
Blocking other people from participating doesn’t protect expertise. Responding well to what they make does.
This article has a footnote
There’s an obvious footnote to everything I’ve just written: this article itself.
AI helped me work on it.
The observation and the argument are mine, along with the experiences that shaped them, the things I disagreed with, the revisions I chose and the decisions about what survived. But I used AI to interrogate the idea, find weaknesses in the argument and improve the writing.
I could have refused that help on principle. The article might have taken longer, come out worse or joined the hundreds of other half-developed thoughts already sitting in my notes.
Instead, you’re reading it.
Of course, AI can substitute for thinking too. I could have typed “write me a thought-leadership article about AI and design,” published whatever came back and contributed little of myself.
That’s the same distinction all over again.
Did the tool help develop the thinking, or did it substitute for having any?
The output might look similar from a distance, but the process isn’t.
And someone will decide that using AI anywhere in the writing process invalidates the argument. In which case, I suppose we’ve arrived back at the beginning.
The two birds
None of this means quality stops mattering, that every unsolicited idea is good, or that everyone should spend half their week moonlighting in someone else’s discipline. The argument is narrower than that.
If someone’s responsibilities are handled, their contribution creates useful information, and they treat it as an input rather than a mandate, more initiative is a good problem to have.
The theoretical process where every artifact originates from exactly the person whose title matches it might look cleaner on an org chart. I’m not convinced it produces better work.
This article, appropriately enough, is a bird in the hand. It started as an observation, became a rough draft and got challenged, rewritten, cut apart and improved along the way, with some help from AI.
Whether I personally typed every word doesn’t make it more or less valuable.
What matters more to me is whether there was something here worth thinking about.
I’ll leave that part to you.
Systems that remember
Why modern design systems aren’t libraries, they’re organisational memory. TL;DR — Key Takeawa
Beyond AAA: The Accessibility Gold Standard Checklist
For years, I treated accessibility as a finish line. Hit Level AA or AAA, check the box, move on. Bu
From Execution to Influence
The Mindset Shift That Elevates Designers Most designers chasing the next level focus on tools, poli