It's Not What You Need, It's What the AI Needs

I was in a cooking class in Italy. Florence. Small kitchen. Flour everywhere. Hands deep in pasta dough.

And it was falling apart.

I was doing exactly what you're supposed to do. Or what I thought you were supposed to do. Measuring carefully. Kneading with purpose. Trying to control every part of the process. My hands were tight around the dough, pressing it into what I wanted it to be.

The dough was crumbling.

The chef (this older Italian woman who'd been making pasta longer than I've been alive) walked over. Watched me for a second. Didn't rush.

Then she put her hands over mine. Loosened my grip. And said something I think about almost every day:

It's not what you need. It's what the pasta needs.

Not what you need. What the pasta needs.

I laughed. Loosened my hands. And verbalized what I already knew: I have control issues.

Apparently it's not just pasta. It's AI too.

The vending machine problem

Here's how most people use AI:

They walk up to it like a vending machine. Insert prompt. Expect output. Get frustrated when the output doesn't match the picture in their head.

"Write me a marketing strategy."

"Create a social media calendar."

"Draft an email to my clients."

And then they're disappointed. The output is generic. Bland. Could be for anyone. So they conclude: AI doesn't work. Or worse, they use the generic output anyway and wonder why it doesn't land.

The problem isn't the AI. The problem is the approach. You're telling it what YOU need. You're not giving it what IT needs.

What AI actually needs from you

It needs context. That's it. That's the whole secret.

Not magic prompts. Not engineering tricks. Not the right model or the right subscription tier. Context.

• Your voice. How you actually talk, what words you use, what words you'd never use.
• Your audience. Who they are, what they care about, what keeps them up at night.
• Your constraints. Your budget, your timeline, your capacity, your non-negotiables.
• Your goals. Not "grow my business" but specifically what success looks like in terms you'd actually measure.

The more of that you give, the better the output. Every single time. It's not a mystery. It's a relationship. The AI performs exactly as well as the context you provide.

And the fastest way to figure out what context is missing? Ask.

Ask it what it needs

Literally. This is the move that changes everything and it's so simple people think I'm kidding.

After you describe what you want, add this: "What additional context would help you do this better?"

Or this: "What questions would you ask me before starting this?"

Or even just: "What do you need from me?"

The AI will tell you. It will ask about your audience, your tone, your constraints, your examples of what good looks like. It will surface the gaps in what you provided. And each answer you give back makes the next output dramatically better.

You're not prompting. You're having a conversation. And the conversation IS the prompt.

It's not what you need. It's what the AI needs.

Start small

The other mistake: going too big too fast.

"Write me a complete marketing strategy for my wellness studio." That's not a prompt. That's a prayer. You're asking the AI to make a hundred decisions it doesn't have context for.

Instead: "Here's my wellness studio. Here are my three services. Here's who my clients are. What would you want to know about my business before building a marketing strategy?"

Let the AI interview you. Let it ask the dumb questions. Let it surface things you haven't thought about. Start with a conversation, not a deliverable.

Then build from there. One section at a time. Each piece informed by the conversation before it. Small, contextual, iterative. Just like making pasta. You don't start with the finished dish. You start with the dough. And you listen to what it needs.

Treat it like a build partner

Not a search engine. A search engine takes a question and gives you links. That's lookup.

Not an employee. An employee takes instructions and follows them. That's execution.

Not an oracle. An oracle knows things you don't. AI doesn't know things. It processes patterns.

A build partner. Someone who has different knowledge than you do, brings a different perspective, and needs your context to do good work. The way I work with clients is the same way I'm telling you to work with AI. Conversational, collaborative, context-rich.

When I sit down with a client, I don't hand them a questionnaire and disappear. I ask questions. I listen. I follow up on things that don't quite make sense. I share what I'm thinking and see if it resonates. The conversation IS the strategy.

AI works the same way. The conversation is the prompt. The relationship is the output quality.

Why I build voice frameworks

This connects to everything else I do.

When I build a voice framework for a client, I'm not just documenting how they talk. I'm building a context package for AI. Every time that client uses AI for content (whether it's through their own tools or ChatGPT or anything else) the voice framework gives the AI what it needs to sound like them.

Without the voice framework, the AI guesses. With it, the AI knows.

That's the whole business model, honestly. I build the context layer that makes AI work for your specific business. A strategic assessment figures out what context is missing. The build creates the tools and frameworks. Ongoing support keeps it evolving as your business does.

It's not what you need from AI. It's what AI needs from you. And I help you build the thing that gives it what it needs.

Back to the pasta

That chef in Florence didn't teach me a technique. She taught me a posture.

Let go of the grip. Stop forcing the outcome. Feel what's actually happening instead of what you think should be happening. The material has its own logic, and your job is to work with it. Not against it.

The pasta came together when I stopped controlling it. The AI works when you stop commanding it.

Not what you need. What the pasta needs. What the AI needs.

Same lesson. Different kitchen.