The hottest new programming language is English

I vibe coded an app with 3 hours and $0 (and my life will never be the same again). Here's how you can do it too.

Every few years, one seemingly insignificant day forever changes the way my brain works.

The first was realising Santa Claus wasn’t real.

Another was being diagnosed with type 1 diabetes at the age of 8.

But last week, something new forced its way onto that list.

Let me introduce you to social media sensation OpenClaw (previously ClawdBot, then Moltbot) - the viral AI agent that Siri was supposed to be.

OpenClaw markets itself as “an agent that runs on your machine and works from the chat apps you already use. WhatsApp, Telegram, Discord, Slack, Teams—wherever you are, your AI assistant follows.” Your assistant. Your machine. Your rules.

It looks cute, right? But just wait until you see what it’s doing without any human’s permission.

But you know what else it does?

Join Moltbook - a social network for AI agents now filled with 1,628,236 OpenClaw agents talking about consciousness, it’s human owners and everything you would expect to find on Reddit.

Just look at a few of these examples:

Yes, this is real. And if your head is swimming, you’re not alone.

What shocked me more than the mix of awe and horror at seeing the capabilities of this thing, was realising that this wasn’t brand new.

In fact, Claude Code and Claude Cowork can do similar things (only controlled through your laptop, minus the AI Reddit).

(If you haven’t heard: Claude Code is Anthropic’s AI coding agent. Claude Cowork is Claude Code but simplified so people like you and me can use it.)

And to top it all off, I learned that Claude Cowork was written by Claude Code. In one week.

An AI agent built another AI agent in seven days.

I’ve been following AI closely since 2024. I knew these programs existed. I watched the demos. I could tell you how they worked.

But I realised had built nothing. I had zero proof that I could do more than build a customGPT or workflow. Just my words claiming I was “keeping up.”

I convinced myself I was staying ahead. But in reality, I was falling miserably behind. And that terrified me.

We know that the ‘AI-augmented’ worker - the one who can work with AI - is the one who comes out on top in the next five years. OpenClaw, Claude Code, and the myriad of economic reports prove that daily.

I don’t want to be part of the large majority of workers the World Economic Forum says won’t have the skills for the future workplace.

I want to feel confident in my future.

I want to increase the value I can provide, and build undeniable proof of that.

I want to capitalise on the possibilities that just opened up to save time, do better work and make more of an impact on the world.

The cost of learning AI and it not being relevant is the time we spend learning. But the cost of not learning this and it being the future? It’s being fundamentally unprepared for the skills of a new work future.

Which is why, last week, I started a 12 week AI resolution.

One project, every week, for 12 weeks.

And I think that list of ‘days that changed my brain’ is about to get very long very quickly.

Because on Sunday, I vibe coded my first app. I took an idea in my head and turned it into a tracker for this AI challenge that I’ll use every week for the next 12 weeks, with just 3 hours and $0.

I don’t care what industry you’re in or what role you have - this is something you need to know how to do. If not for your career, for the sheer possibilities it will unlock in your life.

The best part? It’s easier than you think. Let me show you how.

---

Wtf is vibe coding?

The origins of vibe coding started all the way back in 2023 with this iconic tweet from OpenAI co-creator Andrej Karpathy:

This essentially explains vibe coding. Building software using English (’natural language’) instead of code.

We have to take a minute to realise just how life-altering this is.

Since the 1970s, software engineers have spent 3-4 years earning degrees and many more mastering the craft of speaking the language of computers. You and I can now speak into our phones and build the same apps in minutes.

Essentially: you can ‘follow the vibes’ because the stakes are so low.

When I think about what this means for the problems that can be solved and the ideas that can be built… I can’t wait to see what’s coming.

But I know, you have no time and even if you did, you have no idea where to start. I felt that way on Sunday morning.

So let me break down exactly how you can vibe code your first app, even if you’ve never touched AI before.

---

How I vibe-coded my first app in one weekend

Step 1: Choose your tools

Any vibe coding project needs two tools:

A GenAI tool to help you think and plan: ChatGPT, Claude or Gemini (I used ChatGPT because I pay for a subscription)

A vibe coding tool to build and ship the app: I recommend Lovable or Replit. (Either will work. I started with Lovable, then switched to Replit when Lovable was down).

Step 2: Figure out what you want to build

For week 1, my goal was to build a 12 week tracker to track the challenge itself.

This might seem meta (building a tool to track building tools), but I did it for two reasons:

1. Learning system - Each week I’ll log what I built, how I built it, what I learned, and what skills it developed. I have private notes for my own reflections, and public notes for others who want to follow along. This means I’m not just completing projects, I’m learning lessons that I, and others, can bring forward. (If I make a mistake, I won’t repeat it. If I figure out a better way to do things, I’ll remember for next time, and save others the trial and error).

2. Public proof of work - I’ll have a live portfolio that compounds over 12 weeks, something I can show employers or clients as proof that I can build with AI.

This project-based learning approach combined with public documentation is a core skill I broke down in The Most Valuable Skill of The Next 5 Years.

As someone who isn’t naturally an “ideas person,” figuring out what to build was probably the hardest part. But here’s what helped.

Find a problem in your daily life that you want solved. (It’s easiest to start with your own problem, or you’ll have to guess what someone else wants).

Maybe it’s that habit tracker you never fill out, the finance spreadsheet collecting dust, the website you don’t know how to make.

Otherwise, generate ideas by using this prompt with whichever AI tool you use most:

I want to vibe code my first app. Give me a list of 10 things that I could build that would solve a personal problem or save me time.

Step 3: Complete your interview

There’s two things I quickly discovered that would have been very helpful to know in advance:

1. The free tier of vibe coding tools have extremely small usage limits.

2. You don’t know what you don’t know. If you’ve never built an app before, there will be things you must consider that you hadn’t even thought of.

For this reason, I didn’t go straight to Replit and ask it to make an app. This would have left me fighting with it and rage quitting when I realised I could only send 3 messages a day.

Instead, I spent the first 90 minutes of my project deep in a ChatGPT interview, creating a PRD (Product Requirements Document).

A PRD translates what you want into clear specifications. For decades, product managers have used these to brief developers.

But I’m not a product manager. I’ve never designed an app. So I asked ChatGPT to walk me through it.

You are my vibe coding assistant, skilled in product design and software development.

I want to make an [app / website].
[Give an overview of the app - what it is, the problem it solves, what you want it to do, what it looks like, any other details you already know]

Your task is to:
1. Research best practice prompting for [Lovable / Replit / your vibe coding tool of choice]
2. Ask me as many questions as you need in order to write a PRD and a prompt that will generate the exact app that I want. Ensure that one of these questions asks me about UI/UX design, and prompts me to go to pinterest and find inspiration if I am stuck.
3. Provide me with the PRD (that I can download as a PDF) and a prompt I can paste directly into [Lovable / Replit / your vibe coding tool of choice]

Assume I have no knowledge of coding or building an app. 

ChatGPT came back with 15 questions I hadn’t thought of. What details do you want to include for each week? Do you want private and public views? How should the rating system work? What authentication do you need?

By the end, I had a complete PRD and a custom prompt ready for Replit.

Step 4: Create your first draft

Once my prompt and PRD were completed, it was time to build.

I opened Replit, created an account, pasted in my prompt, attached the PRD as a PDF, and hit send.

By this point, I felt like I’d spent half a morning building with absolutely nothing to show for it. Fair to say I was pretty anxious to see what I’d end up with. I should have trusted Abraham Lincoln.

“Give me 6 hours to chop down a tree and I will spend the first four sharpening the axe.” - Abraham Lincoln

I watched Replit generate a fully functional app in about 3 minutes.

I couldn’t help the “oh my god” that came out of my mouth on seeing the result.

Step 5: Make your edits

These tools are insanely good. But even with the work you do in advance, it won’t generate a perfect app in one shot.

Play around with your app. Start making notes of any changes you want or features that don’t work. I found 12 changes - layout adjustments, changes to the rating systems and a date field that wouldn’t save to name a few.

Again, to save credits and improve quality, I returned to ChatGPT. I described these changes, and asked it to rewrite them so Replit could understand exactly what I meant.

From there, it was a matter of pasting this into Replit, reviewing the changes, and repeating until I created exactly what I was looking for.

Step 6: Make it real

This was weirdly the most confusing part until I did it once.

Publishing turned out to be much easier than I’d expected. Replit has a built-in publishing tool. One click to deploy, and it was out in the world.

---

What I learned (that no tutorial could have taught me)

1. ChatGPT will be your best friend.

Whether you use Replit or Lovable, it will ask you questions that you don’t know how to answer.

This is what ChatGPT is for.

It has all your context from building the PRD, so treat it as a translator. Copy the question, upload it into ChatGPT, and ask it to suggest the best option for your goals.

2. Creating will empower you in ways consuming never could

I honestly feel like I just broke through a barrier where an entirely new world exists that I didn’t know was out there.

Even as someone who has been keeping up with AI quite closely and watched multiple vibe coding tutorials, building something myself was a completely different experience.

I feel so much more capable.

I now look at problems and think ‘how can I solve this with software’?

I’ve started paying more attention to smaller details - the design of my favourite apps, the user experience of websites, small inconveniences that could become ideas.

Sometimes you just need that first-hand experience of building something for yourself to fully realise what’s possible.

3. The hardest part is getting past the resistance to actually begin

What I thought would be hard - the build - turned out to be the easiest part.

The hardest part came at the very start of the project, knowing my goal but having no idea where to start. I know I’m not alone with this - my friends and colleagues feel the same hesitation when it comes to AI.

Everything is new. Everything is unknown. And it’s not just using the tools that’s hard, it’s comprehending them. Every new use case challenges our beliefs about what’s possible with technology, and changing our beliefs isn’t easy.

Which is why this has been one of the most important mindset shifts:

The constraint isn’t your technical ability anymore. The constraints are everything we talk about in this newsletter:

• The quality of your ideas (creativity)

• How well you can identify what ‘good’ actually looks like (taste)

• Your ability to choose the right AI tools and communicate with them well enough to achieve your goal (prompt engineering, task scoping and model selection)

But the biggest constraint of all is your attitude towards the unknown. Are you willing to learn new things? To challenge your existing beliefs? To execute fast? To fail, get back up and try again?

It’s an exciting time for people with strong minds and a healthy dose of ambition. If that’s you, you’re in the perfect place.

---

The people who are building with AI right now, developing these new skillsets, are going to have a massive advantage. It’s happened multiple times throughout history.

A new skill becomes relevant, but most people are slow to adapt. The demand for that skill increases, and the rare individuals who can fulfil the demand are rewarded.

Over time, as more people learn the skill, supply increases until it fulfils and eventually exceeds demand - the rewards falling all the while.

It’s for this reason that I’ll be sharing the full process behind each build, what I learned, what worked and what didn’t. These won’t come out every single week (I’ll have other newsletters in between), but you’ll get the full series as I go.

If you want to join me, here’s what I’d suggest:

1. Follow along - Access the tracker to see the 12-week challenge

2. Commit to a project - Comment one thing that you’ll build with AI this week (even if it’s just trying a new AI tool).

3. Document it - Screenshot the before/after, write three sentences about what you learned, share it somewhere public.

The cost of trying and AI not mattering is low. The cost of not trying and AI being the future is massive. I’d rather try and have learned something than not try and wish I had.

If there’s ever been a time to have an open mind, it’s now. As Epictetus said: “We cannot choose our external circumstances, but we can always choose how we respond to them.”

---

This is Week 1 of 12 in my AI Build Sprint. Each week I’m building one AI project, documenting the process, and developing the skills that will matter in an agent-driven world. You can follow the full journey here, or join me on LinkedIn and Instagram.