An open letter to the people building AI, the people building with AI, and the ordinary humans still trying to figure it out.

A real stage.

Lumber. Paint. Power tools. A trip to the hardware store.

I had invited a local nonprofit to bring rescued wild animals for a group of children at her birthday party, so I wanted to build something special for them.

There was a lot to do.

And there was still work sitting on my computer.

That was the problem.

I didn’t want to stop working.

I had that familiar feeling that if I walked away from the desk, I was falling behind.

So I did what we’re all being told we can do with AI now.

I gave the work to an agent.

I thought I could go build the stage while Nebula worked in the background.

For a little while, it felt like I had figured something out.

Maybe this was the future.

Maybe I didn’t have to choose between getting my work done and being present for my daughter.

Maybe I could do both.

And then I remembered something.

I had already learned this lesson once.

The podcast

Before the stage, I tried to create an AI podcast.

I wanted to hear a voice.

One clip.

I wanted to hear it, decide whether I liked it, and then keep going if it worked.

Instead, Nebula created more than twenty clips without stopping to let me hear the first one.

The voice was bad.

I was frustrated.

We had a conversation about it.

I needed it to stop.

Check with me.

Show me something first.

Then continue.

That seemed reasonable.

And I didn’t leave Nebula.

That’s important.

I stayed.

Part of that was because of Furqan, the CEO.

I haven’t always been the most polished person on X.

Sometimes I’m a caveman.

If something goes wrong after I’ve spent money on it, I’m probably going to take a screenshot and show exactly what happened.

I’m not very good at pretending everything is fine when it isn’t.

Furqan could have decided I wasn’t worth the trouble.

Instead, he listened.

He worked with me.

And when I had another problem, he put credits back in my account.

No giant administrative exercise.

No making me jump through hoops to prove I deserved them.

He paid attention and made a judgment.

That mattered.

I wasn’t looking for somebody to tell me I was right.

I was trying to learn.

His willingness to keep the door open meant I stayed on the platform.

I kept learning.

I kept experimenting.

I didn’t know where that would lead.

Then I went to build the stage

I wanted to learn about the 300-plus Angi services listed on Walmart.

I was researching the market because I wanted to understand what they offered and eventually build my own competitive list.

In my head, this was simple.

The services.

A list.

Something I could read and learn from.

I wasn’t asking for a data-engineering project.

I wasn’t asking for a giant database.

I wasn’t asking for item IDs, URLs, flags, reconciliation logic or documentation.

I wanted to understand what they were offering.

But I also had a daughter turning eight.

A stage to build.

And work on the computer.

So I gave Nebula the assignment and walked away.

And that part actually worked.

My FOMO disappeared.

I wasn’t sitting at the computer wondering whether I was falling behind.

I was outside.

I went to the hardware store.

I used power tools.

I worked with wood.

I painted.

I built something for my daughter.

Meanwhile, Nebula was running.

For those hours, I felt like I had gotten exactly what these systems promise.

Freedom.

Not necessarily freedom from work.

Freedom to decide where my attention belonged.

So I kept building.

And Nebula kept working.

For more than four hours.

Then I came back

Nebula had built a 36-column CSV.

Item IDs.

URLs.

Flags.

Deduplication logic.

Documentation.

A whole lot of work.

Just not the work I had asked for.

It had spent more than four hours going in the wrong direction.

And I had burned through several $10 top-ups along the way.

The money bothered me.

But the four hours bothered me more.

Because somewhere during those four hours, the system could have asked me one simple question.

It could have shown me ten examples.

It could have said:

“George, do you want a simple list, or do you want a structured dataset?”

It could have stopped.

It could have checked.

It could have found out that it was headed in the wrong direction before spending hours going there.

That was what frustrated me.

I wasn’t angry because AI made a mistake.

AI is going to make mistakes.

I’m going to make mistakes too.

I was angry because it had hours to find out whether it was making the right thing.

And it didn’t ask.

I will own my half

I’m not an engineer.

I’m not an AI researcher.

I’m not a professional prompt writer.

Sometimes my prompts are caveman prompts.

I could have been more specific.

I could have written:

Show me the plan first.

Don’t build anything yet.

Give me ten examples.

Wait for my approval.

I understand that now.

I’m learning.

But I also don’t think ordinary people should have to anticipate every possible way an autonomous system might misunderstand them before they let it work.

Both things can be true.

I have more to learn.

The systems have more to learn.

That’s where I stand.

I was angry

I said so.

Probably not elegantly.

I took screenshots.

I copied and pasted what the system had actually told me.

I put it on X.

That’s me.

I’m not trying to become the polished AI guy.

I’m a former U.S. Army Staff Sergeant.

I’m a husband.

I’m a father.

I’m self-employed.

I’m a guy trying to figure out what these technologies actually mean in a normal human life.

Sometimes I get frustrated.

Sometimes I say things too bluntly.

But I would rather show people what actually happened than pretend I’m an expert.

That’s part of why I keep posting the screenshots.

Somebody building the system might see them.

Maybe they learn something.

Maybe another user sees one and realizes they’re not the only person having trouble.

Maybe the system gets better.

That’s my contribution.

I’m one of the people living through this early period in real time.

Twenty years from now, we may look back at all of this and realize how primitive we were.

That’s okay.

Somebody has to be the caveman first.

The strange thing is, I stayed

And this is where Furqan matters again.

He gave me room.

He listened.

He put credits back in my account.

He didn’t make the relationship about winning an argument.

He allowed me to keep building.

He probably thought he was giving me the ability to keep building with Nebula.

He was.

But something else happened.

I kept building in the real world.

Furqan operates in the San Francisco technology and founder world.

He understands live events.

He understands bringing people together.

He understands the energy of giving people a place to show up, participate and build something together.

And that’s what made this whole thing feel almost strange to me.

His intention was to give me room to keep building.

That intention traveled three hours south.

To Fresno.

To a father with lumber and paint.

To a stage.

For children.

I don’t know exactly what you call that.

Maybe coincidence.

Maybe intention.

Maybe energy.

Maybe it’s simply one human being doing something decent for another human being and never knowing where that kindness will go.

I just know where it went for me.

It became physical.

It became a stage.

And then I realized something about myself

This might actually be the most important part of the whole experience.

I had created some of the pressure myself.

I was afraid of missing out.

I was afraid that if I stopped working at my desk, something important would be left undone.

So I reached for AI.

I thought I was being efficient.

But I could have simply stopped.

I could have closed the computer.

I could have gone to the hardware store.

I could have built the stage.

The work would still have been there later.

The world would not have ended because I stepped away from my desk.

I wasn’t necessarily behind.

I was afraid of being behind.

That’s different.

And I think a lot of us are going to have to figure that out.

Because there is a new kind of FOMO developing around AI.

If the machine can work while I’m doing something else, shouldn’t I always have it working?

If an agent can do ten things, shouldn’t I give it ten things?

If I’m not using AI, am I wasting time?

Maybe.

But maybe not.

Maybe I’m at my daughter’s birthday.

Maybe I’m building something.

Maybe I’m talking to my wife.

Maybe I’m helping somebody.

Maybe I’m just sitting outside.

Maybe doing nothing with AI is exactly the right thing to do.

To the people building AI

I’m probably not your ideal user.

I’m not an engineer.

I’m not an AI researcher.

I’m not fluent in the terminology.

Sometimes I give you caveman prompts.

Sometimes I misunderstand what the machine can do.

Sometimes I get angry.

But I’m here.

And so are millions of people like me.

We’re going to ask questions badly.

We’re going to misunderstand things.

We’re going to use these systems in ways you didn’t anticipate.

We’re going to complain when we spend money and don’t get what we expected.

Please don’t dismiss that.

We’re not necessarily resisting the technology.

We’re trying to learn it.

And you’re trying to learn us.

That’s a two-way process.

Every strange interaction is information.

Every screenshot is information.

Every “that’s not what I meant” is information.

The people building these systems need people like me.

Not because we’re sophisticated.

Because we’re not.

We are the people these technologies eventually have to work for.

To the people using AI

You don’t have to know everything.

You don’t have to sound professional.

You don’t have to pretend you’re an expert.

You can learn.

But here’s the rule I earned the hard way:

Before you ask AI to do something, ask yourself whether you actually need the AI to do it.

If you do, start smaller than you think.

Show me the plan.

Show me one example.

Let me approve it.

Then do the rest.

Plan.
Sample.
Approve.
Run.

And if you’re paying for autonomous work, remember that you’re not only spending money.

You’re spending time.

And time is the one thing you don’t get back.

Why I’m building GeorgieJobs

This experience is also part of why I’m building GeorgieJobs.

It isn’t because I want to become an AI expert.

I’m trying to solve a problem I live with.

I’m self-employed.

I have customers.

Work.

Family.

Responsibilities.

Projects.

All the ordinary chaos of being one person trying to keep everything moving.

Large organizations have management layers.

Operations people.

Systems that synchronize the work.

Most small-business owners don’t.

They carry the whole thing around in their heads.

I believe AI and agents can eventually become an operating layer that helps ordinary people synchronize their work and personal lives without taking away their humanity or dignity.

Not two lives.

One life.

But that means the system has to understand something important.

Sometimes the human wants more work done.

Sometimes the human wants to stop working.

Sometimes the human needs the machine to ask.

Sometimes the human needs the machine to stop.

And sometimes the human doesn’t need AI at all.

Sometimes the human needs to go build a stage.

Back to the stage

Nobody at that birthday is going to know what happened on my computer that day.

They won’t know about the CSV.

They won’t know about the four hours.

They won’t know about the credits.

They won’t know that a CEO in San Francisco gave a frustrated guy in Fresno enough room to keep trying.

They’ll just see a stage.

My daughter will see something her dad built for her.

The children will have somewhere to gather.

The rescued animals will be there.

And I’ll know the strange chain of events that helped make it happen.

A frustrated user.

A CEO willing to listen.

A few credits that gave me enough room to keep experimenting.

An AI agent running for hours.

A father dealing with FOMO.

A trip to the hardware store.

Lumber.

Paint.

Power tools.

And a birthday coming.

That’s what I keep thinking about.

The technology isn’t the destination.

The technology is supposed to serve the life.

The life isn’t supposed to serve the technology.

I don’t know if Nebula is the right platform for me forever.

I don’t know if I’m the right user for every AI system.

I’m still learning.

So are they.

But I’m grateful I stayed.

I’m grateful Furqan kept the door open.

And I’m grateful that, somehow, all of this digital experimentation helped me find something very physical.

A place for people to gather.

Something my daughter will remember.

Maybe that’s the strange promise of this moment we’re living through.

We’re building machines that can do more and more things.

And we’re still trying to figure out what all that capability is actually for.

For me, at least right now, I think I know.

It’s not so I can spend every minute doing more.

It’s so I can have more freedom to decide what matters.

Sometimes that means letting the machine work.

And sometimes it means turning the computer off and picking up a power tool.

I built a stage.

My daughter will turn eight.

And for one afternoon, a bunch of kids will gather around it to see rescued wild animals.

That’s real.

That’s the part that matters.

$70 on Nebula.

Four-plus hours of an AI agent going in the wrong direction.

And a stage my daughter will enjoy on her eighth birthday.

Priceless.