Key Takeaways

  • AI generation creates a "microwave burrito dilemma" where 90-second execution seduces teams into accepting flawed, low-grade product output.
  • Traditional product development filtered bad ideas before engineering started; instant generation pushes quality control entirely to post-build editing.
  • Machine slop happens when details lack deliberate intent, such as arbitrary colors or motion that could just as easily be something else.
  • Stripe required 56 distinct iterations on a single event animation to turn a raw AI output into a subtle, lifelike experience.

The Microwave Burrito Trap

Katie Dill eats microwave burritos for lunch when she is in a rush. She pulls a rock-hard frozen brick out of the freezer, puts it on a plate, and runs the microwave for 90 seconds. She goes from starving to eating in a minute and a half.

“I'm willing to overlook some pretty serious flaws,” Dill says. “It's nearly inedible, but I'm so enamored by the speed of execution. The same happens with AI.”

When tools produce full interfaces or working code in seconds, teams stop evaluating the quality of what was built. They celebrate the speed instead. Dill compares this moment to the post-WWII construction boom, when rapid prefabricated techniques created drab "zombie buildings" across cities. When production friction drops to zero, teams build fast and stop looking closely.

Why the Quality Filter Moved Post-Build

Software teams used to have quality filters built into every step before engineering wrote a line of code. Scoping meetings, wireframes, critique rounds, and resource limits forced founders to kill weak ideas early.

Instant generation destroyed that early filter.

“Refuse to confuse done with good,” Dill warns. "The filter is gone. It used to be quality filter was essentially built into every stage of the product development process, even before a project started. But the filtering that used to be throughout the process now needs to happen post build when it is a lot harder to say no."

When a feature is already running on screen, human psychology fights against throwing it away. You feel finished. But shipping the first unedited pass fills your product with what Dill calls "zombie UI."

The Role of the Post-Build Editor

To fix this, teams must adopt the posture of an editor rather than an initial author. An editor treats generated output as raw material, not a finished artifact.

Dill points to an observation from Nibil on why generic machine outputs feel wrong: “One of the things that so offends us about AI slop is the sense that the details don't matter. The cup is green but may as well have been blue. An editor takes accountability for every decision, which is in many ways every pixel.”

At Stripe, this standard showed up during the creation of an event animation. The initial automated generation provided a rough base, but it was stiff and uncanny. The designer did not accept the fast draft.

“So he did it again and again and again,” Dill explains. “56 times, 56 iterations later, he got to something truly beautiful. Now, it's got more realism. It's got a little bit of life and it's subtle differences, but you can sense the care.”

Speed gets you to iteration one. Craft requires the remaining 55.

What to Do With This

Open the last feature or marketing asset your team built using generative tools. Look at every color, transition speed, copy block, and button placement. Ask your designer or engineer why each specific choice was made. If the answer is that the prompt returned it that way, reject the build and mandate at least ten deliberate revision passes before shipping.