Prompt Playbook: Vibe Coding PART 1

I was at a VC event last week focused on AI and VC investing. The room was buzzing with talk about ecosystem plays, enterprise software, and nine-figure fundraising rounds. Fancy suits, impressive slides, and lots of talk about "AI-powered solutions."

And the whole time I was sitting there thinking: Yikes. These folks are missing something massive.

Why? Because the ground is shifting beneath the entire software industry in two fundamental ways:

First, new entrants can now spin up enterprise-grade software and compete with established players like Salesforce or Quickbooks almost overnight. The building limitations that once required teams of hundreds of engineers are collapsing. A small team – or even a solo founder – with the right AI tools can build sophisticated applications that would have been unthinkable just months ago.

Second, and perhaps more disruptive, end users themselves are starting to bypass SaaS products entirely. Tired of that stupidly expensive Typeform subscription? No problem. Just head to your AI coding tool of choice and say "Hey, I want to make my own form builder" and off you go. Sound like an exaggeration? It's not as far-fetched as you might think. I've seen non-technical entrepreneurs build custom solutions in hours that would have previously taken weeks and thousands of dollars.

AND, importantly, it’s getting easier and easier to build.

Nothing encapsulates this transformative trend better than "vibe coding" – the phenomenon we'll be exploring throughout this five-part Playbook series.

Let's get started

What is Vibe Coding?

The term "vibe coding" was coined by Andrej Karpathy (former AI director at Tesla and co-founder of OpenAI) in February 2025. He described it as a new approach where you "fully give in to the vibes, embrace exponentials, and forget that the code even exists."

In practice, vibe coding means collaborating with AI to build software through natural language conversation rather than manually typing out code. You describe what you want – often conversationally or even verbally – and the AI generates, refines, and troubleshoots the code for you.

Like using ChatGPT normally but instead of information as the output you get software.

As Karpathy put it: "I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works."

This is fundamentally different from traditional coding. Instead of needing to understand every line of code, syntax rule, and programming concept, you're focused on communicating your intent and desired outcome. The AI handles the technical implementation.

This doesn't mean the code quality is poor. In fact, the latest AI tools often produce clean, well-documented code that follows best practices. What's changed is who needs to understand that code at a detailed level.

Why This Matters for Entrepreneurs

For entrepreneurs specifically, this shift is nothing short of revolutionary.

Drastically Lower Technical Barriers: You no longer need to be a programmer – or hire a team of them – to build software. This democratises software creation for people with business expertise but limited technical skills. I’ll show you how you can spin up and deploy your own software within a couple of hours.

Accelerated Prototyping and Iteration: What might have taken weeks can now happen in hours or even minutes. This means faster feedback cycles and the ability to pivot quickly based on user feedback. You can go from “oooh it would be cool if…” to a functioning prototype in minutes:

Reduced Capital Requirements: Building an MVP no longer requires significant upfront investment in engineering talent. This changes the economics of launching a software business - it’s not about raising capital upfront and then building. It’s about building and generating revenue.

In March 2025, Jared Friedman of Y Combinator revealed that approximately 25% of startups in their current batch have codebases that are 95% AI-generated. That's a staggering statistic, suggesting we're at the beginning of a fundamental shift in how software gets built. This would have been inconceivable a year prior.

The skinny here for entrepreneurs is this: we are generally great at ideas but need help making them reality, especially for software where we need engineers. Now we can go from idea to a working tool ourselves, fast and cost effectively.

I can hear your brain humming with possibilities already…

The Reality Check

Of course, vibe coding isn't magic even if “any sufficiently advanced technology is indistinguishable from magic!”

It comes with limitations and considerations.

Remember - as Karpathy put it: "I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works."

That “mostly” there is doing some heavy lifting.

What kind of limitations are we talking about

• Complex applications may still require traditional coding approaches or at least a deeper understanding of the code being generated.

• Some tools are just too “big” (the sheer amount in the codebase) for current AIs to be able to work on efficiently.

• AI tools can generate bugs, security vulnerabilities, or inefficient code that requires human intervention.

• For production applications, some level of code understanding is still valuable for maintenance and troubleshooting.

• Not all types of software are equally suited to this approach (yet!).

These limitations are shrinking by the day as AI coding assistants become more capable. What was impossible six months ago is routine today, and the pace of improvement shows no signs of slowing.

But they are worth keeping in mind. This will be important especially when we choose our first project in the next Part - we’re not going to try to build a new Netflix right out of the gate.

Getting Started with Vibe Coding

The best way to understand vibe coding is to experience it firsthand. Tools like Cursor, Replit and Windsurf are making this approach accessible to anyone with an idea and a willingness to experiment.

I personally use Cursor but honestly all the tools are pretty solid! I’ll talk about options later.

Over the next four Parts of this playbook, we'll walk through

1. How to identify the right project ideas for vibe coding

2. Selecting and mastering the tools in the vibe coding ecosystem

3. The step-by-step process of bringing your idea to life through vibe coding

4. Strategies for scaling, commercialising, and evolving your vibe-coded creations

By the end of this Playbook, you'll be equipped to harness this revolutionary approach for your own entrepreneurial ideas.

What's Next?

In Part 2, we'll dive into the ideation process. We'll explore how to identify problems that are well-suited for vibe coding solutions and how to scope your projects for success. This will make or break your first foray into vibe coding, hence spending some time here.

Whether you're a seasoned entrepreneur looking to build faster or someone with great ideas but limited technical skills, vibe coding opens up some really exciting new possibilities. We’ll cover this exciting (and very new) topic over the rest of the week!

Keep Prompting,

Kyle