On this page ↓
I get a lot of emails from people saying Claude disappoints them. They tried it once, got a generic response, and decided it wasn’t worth the hype.
Here’s what actually happened: they asked it a vague question and got a vague answer.
The problem isn’t Claude. It’s that they fed it nothing.
The Prompt Trap
“Write me a blog post about sustainable fashion.” Vague input, generic output. ChatGPT will give you something that could’ve been written by a content mill. Claude will give you something slightly better, but still hollow.
Both tools will generate text that sounds professional and says nothing.
This is where everyone gets stuck. They blame the AI. “It’s too generic.” “It doesn’t capture my voice.” “It’s not specific to my business.” True on all counts. But why would it be?
You didn’t tell it who you are.
The Shift
Two years ago, I was a ChatGPT user typing one-off prompts like everyone else. Then I started teaching myself Claude Code. The manual didn’t say “load your entire life into the system.” But I started doing it anyway — dumping my brand voice guide, my business model, my pricing, my client list, my family’s dietary needs, my therapy framework, everything.
The outputs stopped being generic. They started being mine.
My therapist-website clients get sites that actually sound like therapists, not marketing templates. My Instagram content doesn’t read like Ube MKE is trying too hard — it reads like a mom and two kids who genuinely love what they’re doing. When I ask Claude to help me write a proposal, it sounds like me proposing, not like me copying something from a business book.
The AI didn’t get smarter. I just taught it to work like it knew me.
What “Loading Context” Actually Means
Context isn’t just information. It’s the specific frame that makes your output yours.
The Five Layers of Context
- Voice architecture. Not “write formally” but the actual rules I follow: contractions always, no semicolons, active voice, short punchy sentences for emphasis.
- Who I’m talking to. Not “business audience” but “non-technical women 35+ who are tired of bullshit and want to build real things.”
- Business constraints. I’m solopreneuling with specific margins and specific hours. That shapes what I ask Claude to do.
- Brand specifics. Actual product names, actual pricing, actual energy, our values, how we compete. Not generic stuff.
- Real examples from my work. A before-and-after from an actual project teaches Claude more than a principle ever could.
When I ask Claude to help me write something now, I’m not starting from zero. I’m starting from “here’s who I am, here’s what this is for, here’s how it needs to feel, here’s where it fits in my bigger picture.”
The Before/After
“Write a short email inviting someone to my course.”
Generic. Hollow. Could be anyone’s course for anyone’s audience.
“I teach non-technical women how to build apps with Claude Code. Most of them are 35-50, solopreneurs or freelancers, skeptical of AI hype but curious if it could actually save them time. This email is for people who read my Threads and know my vibe but haven’t clicked yet. They need permission to believe they can actually do this. Use my voice: warm, direct, contractions, specific details over vague promises. Keep it under 150 words. Here’s the course landing page for reference: [details]. Here’s an example of my email voice: [real email I’ve sent].”
The second prompt takes an extra 2 minutes to write. The output is incomparable.
The Quick Context Checklist
Before your next prompt, answer: Who are you? Who's the audience? What context does nobody outside your head know? What does done look like? Add one example of the thing you're making.
Why People Don’t Do This
“It takes too long.”
It doesn’t. It takes 2-3 minutes to write a good prompt versus 20 minutes to rewrite a mediocre output three times. Pick the math that works for you.
“I don’t know what to include.”
Start with: Who are you? Who’s the audience? What’s the context nobody outside your head knows? What does done look like? Then add one specific example of the thing you’re trying to make.
“It feels weird to load so much stuff in.”
This is the right instinct pointing in the wrong direction. You’re thinking “that’s a lot of data to feed the system.” True. But systems are built to handle data. What they’re not built to handle is guessing.
The Real Work
Here’s what teaching people Claude has taught me: the skill isn’t learning to prompt. It’s learning to think clearly about your own work.
The people who get the most out of Claude are the ones who’ve already figured out their voice, their audience, their constraints, their values. The ones who’ve articulated what they actually do instead of vaguely waving at it.
Claude just makes that clarity visible. Fast.
So if you ask Claude for help and get something generic back, don’t blame the tool.
Go back. Load your context. Tell it who you are. Tell it what you actually need. Then watch what happens.
The AI isn’t the bottleneck.
You are.
(And you can fix that.)
How to Talk to Robots
The newsletter. Weekly. AI that makes sense. Marketing that doesn't suck.
You're in. Check your inbox to confirm.