AI Won’t Help You Build a Business. Here’s Why.
If you're building with AI, you NEED to read this
You know what nobody tells you about building with AI?
The hard part isn’t shipping. It’s everything that comes after, when the dopamine of launching wears off and you’re left with the daily grind of actually making something work.
Finding people who understand that reality? Rare.
I’ve been on video calls with where we just complained about building AI apps for an hour straight. Then laughed about it. Then brainstormed solutions. Then complained some more.
That’s how you know you’ve found your people.
We’ve walked similar paths, building AI workflows, figuring out what makes systems actually work, wrangling multiple AI tools into something coherent. We’ve both learned the hard way that getting from demo to production requires more than just prompting well.
But the most important thing I’ve learned along the way: AI can ship code faster than anyone, but it can’t do the captain’s job.
Things like interpersonal conversations. Listening to other people’s problems and giving genuine care. Digesting messy ideas and deciding how to bring them to life. Showing up consistently when users need help. Those are non-scalable skills that you have to own, no matter how good your AI tools get.
Orel has been through more difficult stretches than I have. Twelve products that made zero dollars. Six hundred days of building before something worked. And then the daily grind of actually making WriteStack succeed. He responds to every email, DMs every subscriber, fixes bugs at 2 AM.
He shared his journey before in our Friday builder series. Today he’s back with a raw, honest breakdown of what running a business actually requires.
And I’ll be honest, reading this really pushed me to further refactor my own applications. That’s what the right friends brings. They don’t just cheer. They quietly show you what’s possible when you stop making excuses.
In today’s guest article, I hope some of what Orel has learned gives you the same kind of push it gave me.
Here are some of Orel’s work if you want to dive deeper:
The 5 Steps Framework For Solopreneurs To Go From Idea To Profitable Product
Trust Over Conversions: How He Turned 90% Rejections Into a $10k Creator Tool
AI made building accessible to everyone. But that just means execution is where everything starts to happen.
Here’s Orel’s take on why AI won’t help you build a business, and what actually will.
Hey everybody!
Orel here, the founder of WriteStack.
On August 23’ I quit my job to become a full time solopreneur, and for the past 26 months I’ve been building products and trying to make it on my own.
During my first 600 days I shipped over 12 products, none of which made any money.
After building WriteStack, I finally understood what the real issue was.
It wasn’t my ideas. It wasn’t even execution.
It was the illusion of progress. The feeling that building something fast meant I was actually building a business.
And that illusion has only gotten stronger since AI showed up. Because now, everyone can build something that looks real.
A product, a landing page, a logo. Heck, even an email campaign can now be done automatically with Atlas.
But what it really does it entrap you.
it makes you feel like you’re winning. Like you are making progress, when you haven’t even started playing.
The Mirage
Although AI gives you speed, it doesn’t give you direction.
Yes, you can build an entire app, end to end, in less than a day — tools like Claude Code or Cursor make this reality. After that, the next question is: “ Now what?”
Most people stop there. They are proud of what they built, they are either scared or think it’s not good enough to try and promote it.
And just like that, it dies.
And the easier it is to get to the stage where you have something ready, the more likely you are to drop it.
—
Look, building a business isn’t the code, or the landing page or the fabulous features you have.
Maybe that worked in the 2000s. Not anymore.
A business is that endless loop that starts after you ship your product:
Get people to try your product → Talk to them → Learn + improve + pivot (if needed) → Find more people → …
That’s called execution. And this is work AI cannot do for you.
The problem is, AI gives you the dopamine hit of completion, without the pain of iteration.
And once you taste that false sense of progress, it’s dangerously easy to stay there (trust me. I’ve been there for too long).
The Trap
“AI lowered the barrier of entry, but raised the barrier for success.”
Everyone can build now. And that’s the problem.
When everyone can do something, doing it well suddenly matters a lot more.
Because you are not competing based on how fast you can implement new features, or how beautiful your app looks.
You are competing based on how well you can execute.
Here’s a great way to put it:
AI made speed a commodity.
Execution is the new scarcity.
Because the people who win now aren’t the ones who build the fastest. They’re the ones who stay long enough to make it work.
Now, let’s learn how to execute in a world of speed.
A moment before we start
Let me tell you something, and forgive me for being blunt:
Execution is f*cking boring.
It’s not the “launch day” fun and excitement, or the thrill of hiding in secret a project that’s going to be a killer.
It’s focusing on one product and doing the same thing, over and over, until it stops working.
And then comes the hard part.
Finding out what’s the next move to execute relentlessly on. Then going back to the grind.
Now that I’ve prepared you, let me give you a framework on how to execute.
How To Execute
(In this section I assume that you have an idea you want to start from)
Every product you’ll build will have to go through the same cycle:
Build → Test → Learn → Adjust → Repeat 2-4
That until you start getting traction, low churn rates and not too many bug reports/feature requests.
Let’s unwrap each of these in simple terms.
1\. Build
Make your idea come to life. Build an MVP (Minimum Viable Product) that solves one problem. And before you ship, smoke test it to catch the obvious bugs.
The mindset you should have is:
‘I know my initial idea will not be accurate. So I need to learn why as fast as possible, to improve’
2\. Test
This is the hard part.
This is the stage where you have to find people who’ll use your product and give you feedback. If you are just starting out, you probably don’t have a lot of audience, so getting those people would require you to manually find them.
Be it through content creation, cold outreach (DMs) or whatever other method you have.
You need to get people’s hands on your product. As soon as possible.
3\. Learn
After you find those people, you need to make sure you can learn from them.
You can set up PostHog to view a recording of their usage, or ask them nicely to jump on a call and see in real time how they use it.
You can also ask nicely for feedback afterwards.
Pro tip:
Most people will ignore. Some will say that they’ll try. A tiny portion will try and be cooperative.
For those who said they’ll try, nudge them until they say stop or ignore 3 nudges.
The way that I like to do it is send them a meme. Something like this
4\. Adjust
In this stage you distill everything that you’ve learned in the previous stage and put it into your product.
Be it implementing new features, fixing bugs or pivoting to a completely new direction.
Listen to what your users tell you. Base you actions on that.
How To Execute Like a Beast
In order to be truly good, you must have people trusting you and appreciating you enough to talk about you and share your product.
To do that, you must be reliable and get shit done.
Let me share a quote from Scott D. Clary
The cheat code nobody talks about:
Being reliable. Answer emails quickly.
Show up on time. Do what you said.
Keep small promises. Remember details.In a world where everyone is flaky, reliability looks like genius. It’s not talent.
It’s just doing what you said you’d do when you said you’d do it.
Here’s how I do it:
I turned on notifications from Gmail first, to respond to requests and emails as soon as I see them.
I will set up calls in the middle of the night to help users.
If there’s an urgent bug that needs attention, I’ll drop whatever I am doing and fix it.
I send a DM to EVERY new subscriber of WriteStack, as soon as I see.
I reply to emails and DMs within minutes, if I am not asleep.
I will go over the replays and see if anybody has had any problem or bug. If they did, I immediately send them a message, acknowledging it and fixing it ASAP.
—
In short, I am extremely available and extremely valuable.
Do that for 6 months. You’ll be amazed at the results.
Final Thought
AI made building much easier. But it also made pretending easier.
You don’t have anymore excuses. You can build anything, in record speed time and for under $250.
Here’s a Note from Anfernee that simplifies it:
All your left with is to face what matters most: consistency.
Because once everyone can build, the only edge you have is to work harder and smarter.
Your ability to stay in the loop when it stops being fun.
To show up when no one’s watching.
To keep iterating when everyone else moved on to their next shiny project.
That’s what differentiates the bad from the BEST.
P.S.
If you are a Substack creator who posts Notes on a daily basis, but looks to automate the process, WriteStack is the exact tool for you.
It will let you import, batch schedule and analyze your Notes, so you’re always on top of your game.
Oh, and if you’re into personalized AI, it’s included ;)