AI has made it incredibly easy to move fast. The question is whether we're moving in the right direction.
I was recently speaking with a contractor who told me he believes the quality of roofing installations has declined since nail guns became widely available. In fact, he said he'd only have his own roof installed using a hammer.
Other contractors would disagree. They'd argue the nail gun isn't the problem. It's the person using it. Someone who doesn't know what they're doing can put nails in the wrong places and create leaks much faster than they could with a hammer.
Meanwhile, a skilled contractor using that same tool can install a roof faster, at a lower cost, and with the same quality.
The technology isn't necessarily creating better or worse outcomes. It's amplifying how people work.
The friction we lost
I think we're seeing something similar with AI, but on a much larger scale.
Today, you can describe an application in a few sentences and have AI build something functional almost immediately. You can then keep prompting, adding features, and making changes.
That's an incredible capability. It's also where things get tricky.
In the past, building an application required more deliberate steps. You had to think through the architecture, requirements, and design before writing much of the code. There were natural stopping points where you needed to think through a problem before continuing.
Some of the friction we worked so hard to eliminate was actually helping us.
Now, those stopping points are disappearing. You can keep prompting and iterating without ever really asking whether you're still building the right thing.
That doesn't mean we should slow down or try to design the perfect solution before starting. Iteration is critical. The challenge is knowing when to keep moving and when to stop and think.
Restraint is becoming an organizational capability
This gets particularly difficult when working across teams.
Imagine you're building a simple intake form. You start with a clean design and a clear objective. Then ten stakeholders get involved, each with additional requirements, exceptions, and functionality they'd like included.
In the past, adding those features involved enough effort that it naturally forced prioritization.
Now the response can easily become, "Why can't we just vibe code that into the next release?"
And technically, you probably can.
It's also much easier to say yes than no when you're trying to keep everyone happy.
Before long, your simple form has become a complicated application. You've added functionality, moved incredibly fast, and potentially lost the simplicity of the original design.
This is where restraint becomes more than an individual skill. It needs to be part of how teams operate.
We need to intentionally create the stopping points that used to occur naturally. Clear requirements, design principles, decision-making authority, and a willingness to challenge whether additional functionality improves the outcome.
The objective isn't to recreate all the friction we've removed. It's to engineer the right amount of restraint into the process.
An exercise to try with your team
Here's something worth experimenting with on your next AI-enabled project.
Run a challenger design exercise before committing to a solution.
- Define the outcome. Give three to five team members the same problem statement and a clear definition of success.
- Build independently. Give everyone 30–60 minutes to use AI to develop their own solution or prototype.
- Compare the results. What did each person approach differently? Which designs are simpler? What assumptions were made?
- Challenge the requirements. Use those differences to refine what you're building. Decide what belongs and, importantly, what doesn't.
- Then move forward. Align on a design and move quickly, with deliberate checkpoints to make sure you're still headed in the right direction.
You're still leveraging AI. You're still moving fast. You're just using some of that speed to improve the thinking before committing to an approach.
AI has made it much easier to build things. It hasn't made it any less important to know what you're building, why you're building it, and when to stop.
