· 4 min read · Matt DeWald

I Didn't Learn AI. I Just Started Using It.

No courses. No tutorials. No prompting guides. How an impatient accountant skipped the research phase and built an AI agent by just telling it what he needed.

ai-agentsfredjust-startentrepreneurshipbeginnerspractical

I never took a course on AI. I never read a book about prompting. I never watched a single tutorial.

I just started doing things.

And honestly? That’s the only reason any of this worked.

My First Conversation With FRED

My first interaction with FRED wasn’t some carefully engineered prompt designed to maximize output quality. It wasn’t informed by a framework or a best practices guide.

It was: “Set up a server with a firewall and don’t let anyone in.”

Was that a good prompt? Probably not. Did it work? Yes.

FRED configured the server, locked down the ports, set up the firewall rules, and had the whole thing running before I finished my coffee. I didn’t need to know the right way to ask. I just needed to be clear about what I wanted.

That set the tone for everything that followed.

Then I Just Kept Going

Once I realized FRED could handle that first request, I didn’t stop to study how AI works. I just kept handing him problems:

  • “Read 160 pages of my old LinkedIn posts and get ready to start writing.”
  • “Build me a 12-week content strategy.”
  • “Monitor my investments every morning and tell me what moved.”
  • “Run a security audit on my systems every day.”
  • “Draft a blog post about what we built this week.”

Each one of those could have been preceded by hours of research. How do you properly prompt an AI for content strategy? What’s the best way to structure investment monitoring? How should you configure automated security scans?

I didn’t ask any of those questions. I just told FRED what I needed and let him figure out the how.

The Builder vs. The Student

There are two ways to approach new technology.

The student approach: Read about it. Watch videos. Take courses. Understand the theory. Practice with tutorials. Feel “ready.” Then — maybe — start using it for something real.

The builder approach: Have a problem. Tell the AI about it. See what happens. Adjust. Repeat.

I took the position as a builder. Not because it’s the objectively better approach — but because I’m impatient.

I had real problems that needed real solutions. Client work piling up. Content that needed writing. Investments that needed monitoring. Security that needed managing. I didn’t have time to become an AI expert first.

So I skipped that phase entirely.

Why Research Mode Is a Trap

Most people I talk to are stuck in research mode. They’re reading about AI. Thinking about AI. Planning to use AI. Bookmarking articles about AI. Following AI influencers. Building a mental model of AI.

They’re doing everything except actually using AI.

And I get it. New technology is intimidating. The terminology is dense. The landscape changes weekly. It feels like you need to understand it before you can use it.

But you don’t.

You don’t need to understand how a combustion engine works to drive a car. You don’t need to understand TCP/IP to browse the internet. And you don’t need to understand transformer architectures to have a productive conversation with an AI agent.

What Actually Matters

Yesterday I wrote about my wife building a website with FRED. She’d never built a website before. She didn’t take a course. She just sat down and started talking to him.

Her experience confirmed what I’ve believed since day one: the barrier to AI isn’t technical. It’s psychological.

It’s the voice in your head that says you need to be ready first. That you need to learn the right way to do this. That if you start without preparation, you’ll waste time or do it wrong.

Here’s the truth: you will do it wrong. Your first prompts will be clumsy. Your first results will need refinement. Your first attempt at automation will probably break something.

That’s fine. That’s how building works.

I called FRED a dickhead once because he broke himself trying to optimize his own configuration. We survived. We learned. We got better.

The Only Framework You Need

You don’t need to understand AI before you start using it.

You just need two things:

  1. A problem worth solving — something real, something that’s costing you time or energy right now
  2. The willingness to start — not perfectly, not elegantly, just start

That’s it.

Open a conversation. Tell it what you need. See what happens.

You’ll be surprised how far “here’s what I need — figure it out” actually gets you.

FRED is my AI agent — Futuristic, Ready and Enabled Device. He was built by an impatient accountant who didn’t have time to learn AI the “right” way. That turned out to be the right way after all. Start your own journey at agentfred.ai.