I was talking with a colleague today about how GenAI is rewriting all our playbooks. Not just the technical ones—the organizational ones too. We got onto the topic of all the coaching I used to do with folks in traditional corporate SDLC roles. Product owners, business analysts, architects, technical writers. Each of them trapped in a box someone else drew around their job description.

The question I’d always ask them was simple: “What is it, in your job, that you design?”

It would catch people off guard. Especially the ones who didn’t think of themselves as designers.

Everyone Designs Something

The product owner designs the flow of value—what gets built, in what order, and why it matters. The business analyst designs the shape of requirements, the structure of problems, the language we use to talk about solutions. The architect designs the boundaries and relationships between systems. The technical writer designs how understanding reaches an audience. The delivery lead designs the choreography of a team’s work through time.

Every single one of these roles is fundamentally creative. They’re not executing a script. They’re making judgment calls, weighing tradeoffs, imagining futures that don’t exist yet. They’re designing.

We just forgot to call it that.

The Thing GenAI Can’t Do

Here’s what I keep coming back to: in a GenAI world, the thing we need most is our designs. Not our typing. Not our formatting. Not our ability to remember syntax or look up API documentation. Those are table stakes now.

What we need is the empathy. The human and socio-organizational understanding of what makes our businesses tick and our customers feel seen. GenAI has read a lot about empathy. It can describe it eloquently. But it doesn’t have any. It can’t feel the frustration of a user who just wants to pay their bill and go to bed. It can’t sense when a stakeholder is nodding along but has already checked out.

GenAI knows what it’s read on the topic. That’s good information—genuinely useful. But it’s not enough to spark joy. It’s not enough to put out a fire.

Delight Is Non-Deterministic

I was in a conversation with a design executive recently, and I said something that surprised even me: delight is an outcome of what can only be a non-deterministic process.

Think about it. If you could reliably produce delight through a repeatable formula, it would stop being delightful. It would become expected. Delight, by definition, is the experience of something exceeding or subverting your expectations in a way that feels good.

You can’t automate that. You can create the conditions for it. You can hire people who understand it. You can build organizations that make space for it. But you cannot engineer it deterministically.

And yet, that’s exactly what we’re trying to do when we over-systematize creative work.

Agentic Everything (And Its Shadow)

There’s a tremendous opportunity right now to plug agentic automations throughout our post-GenAI business operations. I see it everywhere—opportunities to take the drudgery out of delivery, to accelerate feedback loops, to surface patterns we’d never have time to find ourselves.

But there’s a shadow here too.

Agentic systems tend toward regression to the mean. They learn from what’s already been done. They optimize for what’s already been measured. Left unchecked, they will sand off every edge, smooth out every surprise, and produce output that is aggressively average.

This isn’t a reason to avoid them. It’s a reason to use them deliberately, with human judgment wrapped around them like guardrails.

The Urgent Work

So here’s the thing I want you to sit with: while everyone’s talking about which tools to adopt and which jobs will be automated, the more urgent work is organizational.

We need to engineer our org patterns to align on feedback. Real feedback—not vanity metrics, not green dashboards, but conversations with actual humans about whether the thing we built made their life better.

We need to build stakeholder empathy into our operating rhythm. Not as a checkbox, but as a practice. When did you last watch a customer use your product? When did your architect sit with a support engineer? When did your product owner listen to a sales call?

If we don’t do this—if we let agentic automation run without human guidance—we’ll build organizations that are efficient at producing mediocrity. Fast, cheap, and utterly forgettable.

What’s Your Design?

So let me ask you the same question I used to ask those product owners and analysts and architects:

What is it, in your job, that you design?

And now the harder follow-up: In a world where machines can produce passable versions of almost anything, what is it that only you can bring?

I suspect it’s something that has to do with understanding people. With feeling the weight of a decision. With knowing when to break the rules because the situation demands it.

That’s the work. That’s always been the work. GenAI just made it impossible to ignore.