Creativity Can Be Trained: What a Veteran's Field Manual Taught Me About Disciplined Creativity

Creativity isn't a gift. It's a practice you can train.

Most people quietly believe creativity is something other people were born with. They watch someone produce unexpectedly good work and conclude that person has a gift they don't. I want to retire that belief, because it keeps capable people from ever starting.

Creativity can be trained. Not through a burst of inspiration, but the way any skill is trained: a constraint to push against, a small rep you repeat, and honest feedback on what you finished. I've believed this for a while. This term, a student in my class gave me the clearest proof I've seen.

A class that can't survive on prerequisites

I teach CSE651, Software Development with Agentic AI, at CSTU. I have the honor of having a few veterans in the room, alongside people from a wide spread of domains and technical backgrounds. That spread is the whole point. A class like this can't survive inside the traditional curriculum, where a long list of prerequisites decides who is allowed to begin.

So I run two tracks side by side. One is technical: the harness, the tools, the way you direct an AI agent to build real things. The other is the human element, what an older vocabulary filed under soft skills. How you frame a problem. How you decide. How you keep going when the work gets hard. For most of my career those two tracks lived in different buildings. Agentic AI is the first thing I've seen that lets them become one whole person.

What one veteran did with a project assignment

Aleh Haiko is one of those veterans, and his deliverable is the reason I'm writing this. I gave a project assignment. He took it far above and beyond anything I expected. Published at his lab, lab.alehhaiko.com, is The Blue Book of UX Directives: a field manual for human-centered systems engineering, organized into 9 chapters, 62 subcategories and 550 directives, from human cognition and interaction mechanics to trust and safety and a closing chapter on AI-mediated systems.

What struck me most wasn't the volume. It was the form. In his preface, Aleh explains that he deliberately wrote it in the doctrine language of the U.S. Army, where he served: terse, absolute, command-form. He took a register he knew in his bones and carried it into a new discipline.

“Directives are written to be met, not debated.”— Aleh Haiko, The Blue Book of UX Directives

That is disciplined creativity in one sentence. The constraint, a rigid and unforgiving format, didn't suppress his originality. It gave it a spine. Five hundred and fifty finished directives is what a strong constraint plus steady daily reps looks like when someone commits to them.

Discipline is outside-in. Habit is inside-out.

Watching his work made me sharpen a distinction I use in my own practice. Discipline, and its military cousin the directive, comes from the outside in: a standard, a deadline, a rule that says do this. It is powerful, and it is how many of us get started.

A habit works from the inside out. It is the same action repeated until it stops needing willpower and becomes an effortless internal drive. The best practitioners I know use both: discipline to begin, habit to continue. You can say it either way. Aleh's field manual makes the first one visible. My own version of the second is a method I call WRITITATION®, built around a daily writing habit, where thinking and writing are treated as one integrated activity.

Writing every day isn't about producing content. It's the rep that trains the muscle. It forces you to find out what you actually think, and it builds a body of your own context that an AI agent can then work from. Done daily, it stops being something you have to push yourself to do and becomes something you'd feel strange skipping.

Five steps to train your own disciplined creativity

1. Choose one constraint. A fixed format, a word limit, a standard every piece must meet. A constraint turns a blank page into a decision you can actually make.

2. Set a daily rep you can finish on your worst day. Small and regular beats heroic and occasional. The goal is the habit, not the output of any single day.

3. Write to think. Put your reasoning into words before you ask any tool for output. It shows you what you believe and gives AI your real context instead of a vague request.

4. Use AI as a sparring partner, not the author. Ask it for options, objections and edge cases, then make the call yourself. You're after more serendipity in your thinking, not a replacement for it.

5. Ship it, then review it. Finish something visible and look back at what worked. Feedback on completed work is what turns repetition into improvement.

Why agentic AI makes this newly possible

For a long time, a student with a rich background but no coding history had to cross a wall of prerequisites before their ideas could become anything real. Agentic AI lowers that wall. When the harness handles much of the technical setup, what's left is the part that was always yours: the question you ask, the standard you hold, the taste you apply, the habit you keep.

That is why I'm so proud to teach an AI harness class with this much range. A veteran's discipline, a founder's instinct and an engineer's craft all show up in the same room, and the tools are finally good enough that none of them has to be left at the door. Aleh's work made me believe it, and confirmed it: creativity, trained with discipline and sustained by habit, is for everyone willing to practice.

If you want to build this practice yourself, follow along. Sign up for my newsletter below to hear first about the next cohort of Software Development with Agentic AI, and to get the notes I write on thinking, creating and deciding with AI.