We’re diving into @erik_sagen’s article today, Delivery is a Trap. He describes a moment of realization when he was stitching together polished screens into clean workflows, testing them with a small group (meant to represent everyone), and then calling that validation… until it hit him that testing a picture is not the same as testing reality.
He argues that somewhere along the way, design shifted from shaping thinking to producing artifacts. Instead of helping decide what to build, designers started delivering finished-looking work after direction was already set. When teams focus on perfect screens and polished decks, they get pulled in late. They react instead of lead.
…and once something looks complete, people stop questioning it, even when they should.
Let’s jump into the discussion
You suggest that influence comes from learning early. Presenting at the end is too late. Rough ideas, fast experiments, and real user behavior build trust. Trust is what gets designers into the decisions that matter.
You argue that if design wants strategy back, it has to stop hiding behind deliverables and start showing up where choices are still open.
So here’s what I’d love to discuss
What needs to change to have design to show up before direction is set, not after?
Excited to dig in with Erik Sagen! He’s touching on a subject that many of you have brought up in our exchanges on design impact.
Great to chat with you today Erik! I’m excited to dig into this topic you laid out so clearly in your article.
Let’s start with the obvious question: AI design workflows. In your view, are they creating more short-term influence problems, or do they actually strengthen design’s role as a strategic contributor?
Hey all. Good to be here and thanks for the invite @Bryan - very cool. Hope we can have a good chat today about how delivery is a trap (cue itsatrap.gif).
So, to answer your question, we have to stop acting like we need permission to be messy.
If you wait for the magic meeting invite to “do strategy”, you’re already behind. You have to start showing up in DMs or channels (whatever you use for comms) with a half-baked sketch or a tough question before requirements are even written. It’s all about trading that BIG reveal for a bunch of small, informal conversations.
It’s a lot harder to be boxed in by delivery when you’re already in the room helping everyone figure out in the problem space.
tldr; It’s about being a partner in the “I don’t know” (throwhands.gif) phase instead of “make it pretty” please.
And that second one, it’s a bit of both honestly? It’s mostly exposing the people who were just making drawings.
If your value is tied up into moving pixels or making drawings, AI is a threat because it IS faster than you. Which then creates a short-term panic. But for the strategy side of things, AI is a massive massive win. It can handle that work so we can spend more time on the what and why.
I also see it as a filter of sorts. It’s focusing us to get better at the thinking part of the job.
If you’re using AI to churn out fifty versions of a bad idea, you’re still in that trap. But if you use it quickly to test a hunch and get to the truth faster, well then, you’re probably building that trust I talk about in the article. It’s a tool. A tool for momentum, but not a replacement for judgement.
Here’s what I’m wrestling with next: if AI makes it easier to move fast, but your advice is to show up earlier with something half-baked, what does that actually look like in practice? Do we hold back on polished prototypes? Do we intentionally keep things rough even when AI can make them look finished in minutes?
If you wait for the magic meeting invite to “do strategy”, you’re already behind. You have to start showing up in DMs or channels (whatever you use for comms) with a half-baked sketch or a tough question before requirements are even written.
So how do you balance AI speed with intentional roughness? How do you use the tools without falling back into the same delivery trap, just faster?
Well, speed shouldn’t be a trap for more speed. If AI lets you build a polished-looking screen in a matter of seconds to minutes, you just saved that time to think, not to hit apply and publish.
In practice though, this means being disciplined about WHAT you show. Just because the tool can add cool-looking shadows and perfect gradients doesn’t mean you should let it. I often tell AI to “give me this in wireframe style” or I’ll purposely use a sketchy type font in its generation like (dun dun dun) Comic Sans or Supermarker (Supermarker - Fontwerk).
Might sound silly, but it’s a visual cue to say “Don’t look at the paint, look at the walls.”
The balance though is simple. Use AI to explore ten different ideas in the time it used to draw one. You aren’t holding back on the tech, you’re using it to find the right direction instead of the best-looking one. If a stakeholder sees a perfect screen, they stop giving you honest feedback because to them it’s already done.
If you make anything look 100% done, you’re basically asking for a binary answer; yes/no. If you keep it at 70%, you’re asking for the feedback you need. Which do you really want?
I’ve always thought it’s a better way to influence- get the stakeholders to own more of the problem.
It can be uncomfortable, though. Many designers feel trapped in the idea that they are “wrong” when stakeholders correct them. Any thoughts on how to reframe that dynamic?
Oh! Okay So, being corrected is actually a win. It means the stakeholder is finally paying attention and putting their own thoughts into the work, rather than just passively agreeing.
When they point out a flaw in a rough sketch, they aren’t grading your work (not like receiving an A, B, C, D), they’re helping you both shape the requirements. I usually just say something like: “That’s exactly why we’re looking at this. Now that we see it, the gap or whatever, how should we fix it?”
You’re shifting the goal from being right to finding the truth. Once you stop treating your drawings like the final exam, the ego starts to melt and you’re just two (or more) people solving the puzzle together.
Yes, I love this. I’ve been thinking a lot about how people talk about craft lately. It feels like there’s a bit of misplaced infatuation with it, and that’s often where ego can get stuck.
In my opinion, the real area for growth is in judgment.
→ Taste is sensing what feels right or off
→ Skill is what allows you to make it
→ Craft is using skill and taste under constraints
→ Judgment is deciding what to do next
Lately I’ve been focusing more on the why than the how.
To the business, you can frame as risk or risk management. I don’t talk about testing user empathy. I talk about how spending a few hours on a messy experiment today could potentially save us six weeks of wasted engineering later.
In my own work, I try to lead with my ears. I’m looking for the off notes in a meeting, the things nobody is saying and using my own judgement to decide if we need a prototype to fix it or just a better conversation to name it.
It’s really about being the person who knows where the problem points are before we start to dig them up.
In design, we often wear ideas and feelings on our sleeves (it’s just innate in the work if you care about it). The real work here is separating your ego, as you mentioned.
How much of your time is spent these days working with others outside of design to clarify directions?
Most of it. If my team and I are just making drawings or pushing pixels, I’m failing.
I’d say 70% of my time is spent talking to product, engineering, other stakeholders. Lots of relationship building too because without trust, none of it works; upstream, downstream. No stream.
If we don’t get these conversations right, the best design in the world won’t save the product. I’d also rather spend three hours talking through a messy idea than three weeks drawing the wrong one.
So true. I want to thank you for sharing your ideas and giving us perspective @erik_sagen. We need more dialogue about the topics you covered! I’ll keep the thread open for others to ask questions. Thanks for the insight!
Really appreciate this reminder. There’s so much behind a well-designed product, understanding the reasoning is goal one. Production-ready visuals seem closer and easier than ever, but the framing of the conversation is still behind the presentation style.
So, being corrected is actually a win. It means the stakeholder is finally paying attention and putting their own thoughts into the work, rather than just passively agreeing.