Hot take: Your prompt engineering strategy is likely flawed. It’s time to address the silent killers of creative output and discover what the top 1% are doing differently.
I see it all the time. Founders, marketers, even engineers, staring at a blank text box, trying to conjure magic from an AI model with a few hastily typed words. They’ve read a few blog posts, maybe watched a YouTube tutorial, and they think they’ve got it figured out. They treat it like a search engine, a vending machine for ideas. And they get back exactly what you’d expect: generic, soulless, and utterly forgettable content.
Here’s the thing. Prompting isn’t a dark art. It’s a conversation. And most people are really, really bad at conversations, especially with a machine that has digested a good chunk of the internet. They're stuck in a 2022 mindset, and it's killing their conversions.
The Uncomfortable Truth About Your Prompts
Let’s be honest. You’re probably making at least one of these mistakes. I know I did. When I was first playing around with early image models—long before Stable Diffusion or Sora were household names—I was basically just throwing spaghetti at the wall. My prompts were a laundry list of keywords: “a futuristic city, flying cars, neon signs, cinematic, 4k, detailed.” The results were… fine. They looked cool, but they had no soul. They didn't tell a story. They were just a collection of pixels that vaguely matched my description.
It took me an embarrassingly long time to realize that the model doesn’t need a shopping list. It needs a director. It needs a vision. It needs a story. I was so focused on the what that I completely forgot about the why and the how.
Most of the prompt “gurus” you see online are selling you snake oil. They’re peddling complicated, jargon-filled templates that are just as likely to confuse the model as they are to help it. They want you to believe that there’s a secret formula, a magic incantation that will unlock the AI’s full potential. There isn’t. The secret is that there is no secret. It’s just hard work, iteration, and a willingness to be a little weird.
From Zero to… Well, Not Quite Hero
I remember one specific project at RemoteTeam. We were trying to create some illustrations for our new onboarding flow. We had a clear idea in our heads: friendly, approachable, a little bit quirky. We spent a week trying to prompt an AI to create them. We tried every trick in the book. We used negative prompts, we adjusted the weights, we even tried to fine-tune a model on our existing brand assets. Nothing worked. The results were either terrifyingly corporate or just plain bizarre.
I was about to give up and just hire a human illustrator (which, by the way, is still a great option for many things). But then, on a whim, I tried something different. Instead of describing the image I wanted, I told the AI a story. I wrote a little paragraph about a new employee, nervous on their first day, and how our software was there to help them feel welcome and supported. I didn't even mention the art style I wanted.
The result was… a revelation. It wasn’t perfect, but it was a start. It had emotion. It had a point of view. It felt like it was created by someone who actually understood what we were trying to do. From there, we were able to iterate and refine, and we ended up with a set of illustrations that we were really proud of. It was a lesson I’ve never forgotten. You can read more about some of my other early struggles in my post about my biggest investment mistake.
What the Top 1% Are Doing Differently
So what’s the secret? It’s simple, really. The best prompt engineers aren’t engineers at all. They’re storytellers. They’re psychologists. They’re artists. They understand that the AI is not a tool, it’s a collaborator. And like any collaborator, it needs context, it needs motivation, and it needs a clear creative brief.
I’ve been fortunate enough to invest in some of the biggest names in AI, including Anthropic, OpenAI, and Scale AI. I’ve seen firsthand how the best teams in the world are approaching this stuff. And I can tell you, it looks nothing like what you’re seeing on Twitter. They’re not just writing prompts. They’re building entire narrative frameworks. They’re creating character backstories. They’re thinking about the emotional arc of the content they’re creating.
They’re also not afraid to get their hands dirty. They’re constantly experimenting, constantly pushing the boundaries of what’s possible. They’re not just using the models, they’re breaking them. They’re finding the edge cases, the weird quirks, the unexpected behaviors. And that’s where the real magic happens.
One of the most successful companies in my portfolio has a team of "creative technologists" whose only job is to try and get the AI to do things it’s not supposed to do. They’ve generated some of the most innovative and successful marketing campaigns I’ve ever seen. And it’s all because they’re not afraid to color outside the lines. You can see a similar ethos in how I approached building my career, which I talk about in How I Became the #1 Angel Investor.
Stop Tinkering, Start Directing
So, how do you actually do this? How do you go from being a prompt tinkerer to a creative director? It starts with a shift in mindset. Stop thinking about what you want the AI to make, and start thinking about what you want the AI to feel. What emotion are you trying to evoke? What story are you trying to tell? What’s the one thing you want your audience to remember?
Here are a few practical things you can try:
- Write a creative brief. Before you write a single prompt, write a one-page creative brief. Who is your audience? What is the key message? What is the desired tone of voice? What are the creative constraints? This will force you to think through your ideas before you start generating content.
- Give the AI a persona. Instead of just writing a prompt, give the AI a role to play. "You are a grizzled old detective, cynical but with a heart of gold." "You are a wide-eyed ingenue, full of hope and optimism." You’ll be amazed at how much this can change the output.
- Use sensory language. Don’t just describe what’s happening. Describe the sights, the sounds, the smells. The more vivid you can make your prompts, the more vivid the results will be.
- Embrace the happy accidents. Sometimes the AI will give you something you didn’t expect. Don’t be so quick to dismiss it. It might be the best idea you’ve had all day. I’ve stumbled upon some of my most successful investments by following a path I didn’t initially intend to take.
I get it. This all sounds like a lot of work. And it is. But the alternative is to keep churning out the same generic, soulless content as everyone else. And in a world where everyone has access to the same tools, the only way to stand out is to have a unique point of view. Your prompts are the key to unlocking that.
Look, I’m not saying that prompt engineering is the only thing that matters. You still need a great product, a solid marketing strategy, and a little bit of luck. But in an increasingly crowded market, your ability to communicate your vision to an AI collaborator could be the one thing that sets you apart. Don’t be the founder who gets left behind because you were too busy tinkering to learn how to direct.
It’s time to stop treating AI like a vending machine and start treating it like a creative partner. Your business depends on it.
Frequently Asked Questions
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
How has this view evolved over time?
My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.