What I’d tell someone smart who hasn’t found their way into AI yet
April 5, 2026 · Gev Marotz · 3 min read
A lot of smart people are getting stuck in the same place right now.
They think the first step is to understand AI.
So they open a course. They save a bunch of threads. They watch videos. They try to get the whole thing in their head before they touch anything.
I think that slows people down.
Especially if you’re already good at your job and just haven’t figured out where these tools fit yet.
The better move is usually simpler.
Start with the work.
Pick one thing that would actually help.
A landing page. A proposal. A rough prototype. A workflow you keep repeating.
If you’re fractional, you already know this instinct. You don’t study a client’s world for six months before taking an engagement. You start with the brief.
Better yet, pick something you actually want to improve.
Then use the tools on that.
Not to become an AI person.
Just to get moving.
Why this works
Once you get one useful result, the whole thing feels different.
It stops feeling abstract. It starts feeling usable.
And the questions get better.
You stop asking, “How does AI work?”
You start asking:
How do I structure this page so it makes sense faster? How do I turn my thinking into something I can show someone? How do I get from research to a point of view faster? How do I make this deliverable better without losing a whole day?
That’s a much better place to learn from.
The work gives you the questions. The friction shows you what matters. Your mistakes will usually teach you faster than another video or prompt list.
A simple way to start
This is the basic flow:
1. Start with the problem and the context
Research the problem first.
What are you trying to improve? Who is this for? What is not working right now? What would a useful first version look like? What are strong examples? What might you be missing?
You do not need the perfect prompt.
You just need enough context to get a first pass going.
2. Turn that into something you can react to
A rough outline. A wireframe. A draft. A simple mockup.
Anything that gets the idea out of your head and into the open.
3. Refine it
What feels off? What needs to be clearer? What should change? What should get simpler? What should be removed?
That’s where the better questions usually start.
4. Move it into the environment you actually use
Once the direction is real, move it into whatever environment your team uses.
A shared doc. A slide deck. Your design tool. Your codebase.
Whatever your team actually ships from.
A lot of people get overwhelmed because they try to solve the whole workflow at once.
You usually do not need to.
You just need the next usable version.
If I were starting today
Pick one real project. Something you’d actually use.
Then do this:
Write down the problem in one sentence.
Ask a tool to help you think it through. What’s the best approach? What should be on the page? What are you missing?
Ask it to turn that into a rough first pass. An outline, a draft, a simple structure.
React to what comes back. What is off? What is missing? What needs to change?
Make one round of improvements.
Show it to one other person.
That is enough for the first afternoon.
You do not need to understand the whole field before you start.
You just need something real to work on, a first version to react to, and the willingness to fix what is not working.
Thanks for reading, Gev
P.S. I’m looking for a generalist growth marketer in Canada for a project in the creator and professional tools space. You’ve actually grown something: a niche app, a newsletter, a community. You already use tools like Descript, Claude, or CapCut in your work. Bonus if you’ve worked closely with a design team and know how to brief creatives without stepping on their toes. My DMs are open.
