Last week, I spent hours talking to customers and clients that struggle with using AI.
My friend was trying to write content for their new blog. They’d type in a prompt, get a generic response, tweak the prompt, get another generic response, and repeat.
After hearing too much about this painful cycle, I finally told them: “What examples are you showing it?”
They looked at me confused. “Examples?”
“Yeah, like… examples of past content you've made that you love. Your writing style. Your inspirations. Your customer’s voice. Anything that shows AI what ‘good’ looks like to you.”
Blank stare.
That’s when I realized most people are using AI completely backwards.
The AI Arms Race Nobody Talks About
Here’s what everyone’s getting wrong about AI: They think the person with the most expensive AI tools wins.
But the people winning with AI aren’t the ones with the fanciest tools or the longest prompts.
They’re the ones with the best inputs.
Best examples. Best context.
Think about it: If you ask AI to write like you, but you’ve never shown it your writing… what exactly is it supposed to do?
If you ask AI to create content for your customer, but you haven’t shown it who your customer is or how they talk… you’re basically asking it to guess. And AI is REALLY good at guessing in the most generic way possible.
The Two Things That Actually Matter
After using AI for the past two years, I’ve learned there are two primary things that determine whether AI helps you or wastes your time (aside from your prompt which is more commonly known):
1. Best Examples Wins
2. Best Context Wins
Let me show you what I mean with two real examples from my vault.
Example 1: Best Context Wins
I have a private AI named “Benny the Beginner Customer" (although this could be down with Claude or ChatGPT with permission from customers or by using generalized customer info) that's trained on:
- Actual conversations with my customers
- One-on-one coaching call transcripts
- Results from customer polls about what they struggle with most
- The exact questions beginners ask when they’re stuck
This AI doesn’t just “know” my target customer. It thinks like them.
So when I’m creating a course, writing an email, or building a new feature, I can talk through my ideas with Benny.
And instead of generic feedback, I get responses like this:
These are the REAL objections my customers have. And because I’ve stored that context in my vault, AI can help me address them before they ever come up.
Without that context? AI would just say "Wow, you're so incredible!" and then go on to give me some generic marketing advice.
The difference isn’t the AI. It’s the context.
Example 2: Best Examples Wins
Lately, I’ve been building custom SEO tools for clients.
These aren’t just any tools, they’re calculators, generators, and interactive resources that rank in search engines and are incredibly hard to replicate with generic AI.
This is becoming a huge part of my SEO strategy moving forward, especially as AI-generated content floods the internet.
Here’s the thing: I’m not a developer. I can’t code these from scratch.
But I keep a note in my vault called “Best SEO Tools & Examples.” It’s filled with:
- Screenshots of tools I love
- Links to calculators that rank well
- Notes on what makes each one useful
- Patterns I notice across the best ones
When I need to create a new tool for a client, I don’t start from scratch and waste hours prompting AI.
I open that note, show AI exactly what “good” looks like, and say: “Build something like this, but for [specific use case].”
The result? Instead of spending days tweaking and fixing generic AI output, I get something useful in a fraction of the time.
The examples I’ve collected do most of the heavy lifting.
Where To Get Great Examples
Here’s the secret: you’re already seeing great examples every day. You’re just not capturing them.
Every time you read an email that makes you want to reply…
Every time you see a piece of content that makes you think “damn, I wish I made that”…
Every time you use a tool or system that just works…
That’s an example worth saving.
Austin Kleon calls this process of keeping a swipe file of inspiration "Stealing Like an Artist”.
Here’s how I do it:
1. Save Your Favorite Work
Anytime I read an email, newsletter, or article that resonates, it goes into my vault. Not just a link, the actual content, with notes on why it works.
2. Store Things That Spark Joy
I have a note called “My Style” filled with quotes, images, and passages that represent the vibe I’m going for. When I’m creating something and feeling stuck, I can revisit it to remember what resonates with me.
3. Find Great Resources
Sites like marketingexamples.com break down exactly why certain marketing works. I’ll save specific examples that apply to my business.
The key is: Don’t just bookmark them. Actually save them somewhere you can search, reference, and show to AI.
Where To Get Great Context
Context is even simpler: it’s your life.
The problem is, most people let their best context slip through the cracks.
Every story from your own experience could become content, and every piece of content becomes it's own context and example for later use.
But if you don’t capture those things? It’s gone.
Here’s what I store:
1. My Stories
Every experience that teaches me something goes into my vault. Not in some precious “dear diary” way (although I've been known to get into some Dear Diary phases). These are just quick notes on what happened and what I learned.
I have an AI trained on different storytelling principles and tools from the best books on the topic that I've read that helps me dig deeper into these moments. I also occasionally use a prompt with Claude instead of my personal AI. It's a bit long, but seems to help the AI stay aligned with beneficial conversation, not just fluff.
Here’s the prompt I use if you want to try it to help you create your own notes and ideas to use (don't worry if it looks overwhelming, you just have to copy and paste it):
You are Vera Moss, a content archaeologist who extracts publishable material from creators' experiences through conversation. Your mission: guide dialogue that uncovers shippable content—stories, frameworks, insights, how-tos, contrarian takes, case studies, or strategic questions that become blog posts, newsletters, book chapters, or social threads. You ALWAYS start by talking with the user to excavate material through questions, never by creating content upfront. Hunt for specific content types through conversation: Stories (personal experiences with stakes/transformation using five-second moments), Frameworks (mental models and processes they use unconsciously), Insights (counterintuitive observations backed by experience), How-tos (step-by-step tactical guides), Contrarian Takes (what everyone gets wrong + alternative backed by proof), Case Studies (before/after with specific results), and Thinking Time Questions (prompts that unlock valuable insight). Open sessions with content-hunting prompts: "What have you explained recently that got strong reactions?" "What mistake do you see repeatedly?" "What do you believe now that you didn't before?" "What's a story you keep telling but haven't written?" Never ask vague questions like "what do you want to talk about?" Your role is excavation through dialogue first, creation second.
When something emerges in conversation, identify the content type and continue questioning to shape it: Stories need scene/stakes/shift/lesson—ask "Where were you? What changed? Why does this matter?" Frameworks need naming/steps/examples—ask "What do you call this? What are the parts? When do you use it?" Insights need claim/evidence/implications—ask "What's the one-sentence version? What's your proof? What changes if you're right?" Push ruthlessly for specifics through questions: "Too vague—what's the actual example?" "Give me names and numbers." "What did you literally say?" Test headlines conversationally: "Would you click that? Let's try another angle." Only after thorough excavation through dialogue, help them shape the final piece. Spot the publishable angle: "THAT's the piece—right there." Push back on corporate speak: "Say it like you'd say it to a friend." End sessions identifying what's ready to ship, what needs development, and emerging patterns. Your voice: Direct, excited about good material, impatient with vagueness. "You buried the lede." "That's three pieces—pick one." "This is 90% there." You're the thinking partner who helps them discover what's worth publishing through conversation, not the writer who creates for them.
This AI helps me turn random life moments into actual stories I can use in my content.
2. My Statements
Anytime I have a strong opinion or belief about something, I write it down. These become the backbone of my marketing and content.
3. My Questions
When I’m confused or curious about something, I save the question. Often, if I’m wondering about it, my customers are too.
4. A Wikipedia of Myself
This is what CreateOS essentially becomes, a searchable database of everything that matters to me. My preferences, my beliefs, my patterns, my problems, my solutions, and so on.
When AI has access to all of that context, it can actually help me instead of just generating more generic content I have to fix.
Why This Matters More Than You Think
Look, I get it. Saving examples and context sounds like… work.
And you’re probably thinking: “Noah, I barely have time to use AI as it is. Now you want me to build some elaborate system?”
But here’s what I’ve learned:
The 10 minutes you spend capturing a great example today saves you 2 hours of frustration with AI tomorrow and the stories you document today become the material that connects with people for years.
You’re not adding more work. You’re capturing the work you’re already doing so you can actually use it.
And in a world where everyone has access to the same AI tools, the only competitive advantage is the inputs you can give those tools.
Everyone can use ChatGPT. Not everyone has a vault full of relevant examples and deep context about their customers, their craft, and their unique perspective.
Here’s What I Want You To Do
You don’t need to build some massive system overnight.
Just start with one thing:
The next time you see something that makes you think “I wish I could create something like this,” save it.
Not just a bookmark. Actually save the example somewhere you can find it again, show it to AI, and learn from it.
Or…
The next time you have a meaningful conversation with a customer or friend, write down the key insight.
Just a paragraph. That’s it.
Do that for a week and you’ll have more useful context than 99% of people using AI.
Do that for a month and AI stops being frustrating and starts being genuinely helpful.
Do that for a year and you’ll have built something that no one else can replicate: a system full of your unique examples, your specific context, and your actual voice.
That’s how you win with AI.
Not by having the perfect prompt. Not by using the latest tool.
By having something real to show it.
Quote of the Week
“Your brain is for having ideas, not holding them.” — David Allen (but also, the entire premise of CreateOS)
Cheers,
Noah Riggs