Exploration Used to Be Expensive
The most important business and technology decisions I’ve faced have had the same constraint: exploration was expensive. You could analyze one or two directions deeply, but eventually you had to pick and commit.
So we learned to commit with conviction on limited information. We studied the options, debated the trade-offs, drew diagrams, and made our best call. Sometimes we were right. When we weren’t, we’d already invested in the wrong direction.
AI changes that constraint. Not by making decisions for you, but by making it cheap to explore before you decide.
Beyond the Mockup
Design decisions usually start the same way. A designer creates a few comps — maybe two or three variations — and the team picks one. The variations are limited because each one takes real time to produce. So you’re choosing between a small number of options, and you’re choosing based on static images.
I recently tried something different. Instead of waiting for comps, I used an AI coding tool to build an interactive experience that let stakeholders explore design directions themselves — switching between styles, trying different user profiles, even comparing approaches side by side. I had something working by my second cup of coffee. More like a design playground than a polished product.
The conversation completely changed. Instead of choosing between two or three options that took days to produce, stakeholders were reacting to a dozen directions that took hours. Preferences surfaced faster and dead ends got eliminated earlier. The final direction had real conviction behind it — not just the least objectionable option from a short list.
What the Documentation Didn’t Say
The design side is one thing. The integration side is where assumptions get really expensive.
When you’re building against a third-party API, verifying what it actually gives you used to be real work. You’d read the documentation, trace through the endpoints, maybe write some test calls. But building enough of your application against it to see how it behaves with your specific use cases? That was development effort, not research. So teams did their due diligence, made reasonable assumptions, and moved forward — knowing they might hit surprises once they were deep enough in to feel them.
Now that cost is almost gone. One of my team members recently wired up a third-party API on day one of a new project — not as a sprint task, just to see what they were actually working with. Pulled real data, mapped it against what the design needed, and surfaced the gaps before anyone had written a line of production code.
The data was structured in ways that weren’t obvious from the documentation, and getting what the design actually needed required additional calls and transformations nobody had accounted for. That discovery would have happened eventually — but it happened before any plans were built around assumptions that didn’t hold.
So What Changes?
These aren’t special cases. The same pattern shows up in technology decisions, vendor evaluations, even how I approach writing. The common thread is that exploration was expensive enough to ration. You’d pick your one or two best guesses and go deep. Now you can go wide first, then go deep on what actually holds up.
That doesn’t mean exploring everything. It means fewer questions have to wait until you’re deep into the work. Questions that used to get answered with “we’ll figure it out when we get there” can get answered now, before the stakes are high.
The Part AI Can’t Do
Exploration generates information. Lots of it, fast. But information isn’t insight.
Knowing that a third-party API structures its data differently than you expected is useful. Knowing what to do about it — whether to build a translation layer, push back on the design, or find a different vendor entirely — that requires understanding your constraints and your team. Context that no tool has access to.
The value of cheap exploration isn’t that it makes decisions easier. It’s that you can see what’s ahead before you’ve committed to the path.
Don’t Get Lost Out There
The obvious danger of low-cost exploration: when you can always check one more option, you might never commit. I set boundaries before I begin — what questions am I trying to answer, and what does “enough” look like.
Explore freely. Decide wisely. Those are different skills — and only one of them is new.