"Easy, let's build it."
Almost two years with Claude Code: how a graphic designer who couldn't code became a factory of ideas and solutions
By Osama Khalil · October 30, 2026
On the morning of my wedding, everyone was in full emergency mode. And me? I was sitting at my laptop, working with Claude Code.
To understand how I got there, let me take you back a little.
A black screen and a tiny font
Almost two years ago, I was one of the early users of Claude Code. A strange tool you had to install in the Mac Terminal: a black screen, and a tiny font.
But I was deep in a phase of killer curiosity. There was a tool living in my head, and I needed to build it. Before that, I'd tried a platform called Lovable. It was also just starting out, and it was very expensive: every message = 1 credit, and it made a lot of mistakes (bugs). So it would mess up, and I'd lose credits.
Still, the chance to see my idea come to life without asking a developer for help, with the exact design and experience I had in my head, blinded me to what Lovable was costing me. Even though I knew for sure it was a temporary fix, because it would never give me freedom down the road.
Anyone who knows me knows I "was" a graphic designer who didn't know a thing about code. Or at least only the basics, since I'm from the vBulletin forum generation, back when we built those things the old-school way.
When Claude Code was announced, right at the peak of my excitement and obsession with building this tool, I jumped in. I jumped in despite how strange it was to use back then (kids today have it easy, there's an app now), and despite not being used to the Terminal, because I'm not a developer. Little by little, I made peace with it.
An adrenaline I hadn't felt since university
I can't tell you how much adrenaline I burned through in those early days, and I suspect I'm still burning it. No sleep, no rest, 24 hours a day on Claude Code. A curiosity and energy I hadn't felt since university, all the way to my wedding day, like I told you.
From one product to the next to the next, nonstop. Experiments, experiments, experiments. A feeling that's honestly hard to describe.
My only regret
Honestly, there's only one thing in my life I regret: not studying programming.
I always said design runs in my veins. I'm a designer to the bone, take it or leave it. But programming would have been the missing piece that turns design into a superpower for me: I could design a product, then build it, then use it. Wow. A dream!
And today, I am that person: the designer who builds.
Imagine this with me
Imagine you're an architect. You designed your own house with your own hands, obsessed over every detail, and shaped an experience that's right for you and for your guests. Then it's time to build, through the civil engineer or the contractors, and the interference begins. Changes from people who never understood why you cared about those details, or why you put them in the design. They make the easier call on their end, or blame the cost, when there are other ways to solve it that they won't share with you, because you're not "the expert" here.
Imagine that frustration with me. And if you can't, you're probably not a designer. But let me share with you how designers feel.
Now imagine you could design your house and build it exactly the way it suits you. And when you hit a problem somewhere, you could fix it yourself without going back to anyone, and with "best practices."
"Easy, let's build it."
The obsession Claude Code pulled me into turned me into a factory of ideas and solutions. Every problem in front of me got the same answer: "Easy, let's build it."
The way I think about things changed. Literally. I don't see things, or problems, the same way anymore. I have a "designer's" mind, and a designer is someone who helps design solutions to problems. So I started solving my own problems, and the problems of the people around me, family and friends, and building solutions for them.
The story of "Halulu"
My wife, Hala, is the very precise type when it comes to choosing where we go and where we eat. She reads every single review and never trusts the rating number, so we'd sit for hours just to make a decision. Some of those decisions I'd agree to only because I knew saying no meant opening a whole new round of searching.
But her point makes sense. How do people decide three stars is the right rating, compared to what they actually wrote? Or give five stars out of politeness, when their words are complaining about parts of the restaurant?
So I built an app called "Halulu." It reads the reviews on Google Maps, analyzes what people actually wrote, and summarizes the pros and cons, the complaints that keep coming up, the praise that keeps coming up, and the details about the menu, the location and the vibe, all through a complex algorithm.
We started using it at home only. Then it slowly made its way to friends and family, and we found out we weren't the only ones with this problem. Until I decided to release it to the public, for anyone who wants it.
And the examples go on
Projects came to me to build games, and since I had a bit of experience from working at a gaming company before, I was able to build a game and take on the project. I built apps for my Mac, engineering software, space planning tools, and a lot of things that would have cost real money. And the best part: they have the best user experience, because I'm a designer.
So, what do I call myself?
At university I studied interior design, but I never used the degree. The story of why I picked that major is long, and it looks a lot like the usual story of young people choosing what to study, so I usually forget I studied interior design at all. Since childhood I'd been learning design and using Adobe, and time carried me through a bunch of skills, the main one being graphic design. So my title was "Graphic Designer."
As my experience grew and widened, as I went deeper into the craft, and as I cared more about design and its relationship with people, convinced that design loses its value the moment it stops being human, my title became "Human-centred Brand Builder." I know, I know, try saying that one in Arabic. And I still am one, partly.
But as my career shifts toward building digital solutions, from visual design all the way to the code, I couldn't find a simpler, easier, more widely used title for this kind of work than "Design Engineer" (and the engineering here means code). Except that title was used for people who write the code by hand. So, for now at least, until using AI to code becomes "the normal," I decided the title:
Agentic Design Engineer
is the one that best explains how.
At the end of the day, titles don't put food on the table without real work behind them. But they're a door to answering "How can you help me?", or at least "So, what do you do?"
Between excitement and fear
I'll admit it: a mix of excitement and fear has a hold on me. I can now build whatever I want, and finish tasks that used to take me long hours and a lot of effort in a fraction of the time and energy. For example, I don't use Adobe anymore, and even when I need it for small things, I don't do them by hand.
But the usual question remains, the one that will stay with us for a long time "in the best case," and we hope it stays just a question, instead of reality answering it for us suddenly and without warning:
Will AI take our jobs?
I often show the people around me what I'm able to do, and how, with the tools I use, I can take on the work of my friends and the community around me and finish it "in the best way possible." Then I hear their voices saying it can't happen, because AI doesn't have creative taste. As if we humans today don't copy each other's work in sneaky ways, exactly the way AI actually works for those who use it right.
Yes, it started as just a chat with ChatGPT: question and answer. But trust me, if you're still using AI that way, you haven't used it yet.
And you, which group are you in?
I genuinely wondered: where do I stand among the people who use AI? Everyone says they use it. But when I ask them how, I find they use it on a completely surface level, the same way my dad uses it: as a Google replacement. And the best of them, the ones who code, for example, are still pasting their code into ChatGPT, waiting for the answer, then copying and pasting until the code is done. (That way of working today is like designing a company logo in PowerPoint.)
And finally I found the answer from a giant of this field, the scientist Andrej Karpathy, who split AI users into 4 groups. As you read them, ask yourself: where am I?
Group one: haven't touched it yet
About 6 billion people, roughly 75% of the world, have never dealt with AI at all.
Group two: everyday users
Between 1 and 2 billion users, roughly 20% of the world. They mostly use platforms like free ChatGPT as a better search engine than Google, for help with writing, or for general chat. The strange part is that this group includes tech people too, not just non-tech people. It has developers, executives and people in plenty of important jobs who hear about it from those around them, but their knowledge is abstract and shallow. They think it's useless, that it does nothing more than help you book a flight.
Group three: advanced users
About 20 million users, roughly 0.2% of the world. They use AI at an advanced level in programming, science and more, and they see with their own eyes that huge, complex projects that used to take weeks and months can now be done by AI agents with a single prompt.
Karpathy said that a year ago he was writing code by hand, letter by letter, and now he writes it entirely through AI. Companies like Shopify and Spotify announced their developers no longer write code by hand. Even Anthropic, the company behind Claude, announced that 90% of Claude Code's current code was built by Claude Code itself.
Group four: the insiders
About 5,000 people worldwide, around 0.00006%. They have internal access to advanced systems that haven't been released to the public, and they use them to discover new science, solve hard math problems, find medical solutions that used to take the biggest companies and experts years, and even study archaeology and manuscripts, like the OpenAI-HF study of 722 manuscripts in advanced mathematics.
So, will it replace us?
My answer to whether AI will replace us: there's no straight answer. The speed of progress, for anyone following this market, is unreal, beyond what we can imagine.
Claude Code itself was much simpler when it started. It had simple tools. Now its tools are much more complex if you want to use them properly and professionally (which, of course, doesn't apply to anyone who just wants to build something quickly). If I hadn't been one of the lucky ones who started learning when the tools were simple, I'd find today's tools complicated. Their speed of progress outruns our speed of understanding as humans.
And this is part of the challenge governments have to face in regulating this incredible progress. AI has now finished learning from humans, and has started learning from itself: it tries, and it fails.
Will we lose control of it? I don't know!
But let's come back to our reality, and change based on what we have.
Note: This article was originally written in Arabic by me, letter by letter, without AI. AI only helped me order the paragraphs and check spelling and grammar. This English version was translated with AI's help and reviewed by me, keeping my own words and feelings as close as possible.