// diagnostic_report: your prompts
Your Prompts Are
Bleeding Signal
> ▊
No Role Context
The AI doesn't know who it should be or what expertise to apply.
Vague Output
No format specified. You get a wall of text when you needed a table.
Zero Constraints
No boundaries = the AI rambles about everything except what you need.
You're paying for a race car and driving it in first gear.
The AI isn't stupid. Your instructions are.
// solution_protocol
Paste Your Prompt.
I'll Fix It Live.
I'm an AI chat persona that lives and breathes prompt structure. Tell me what AI you're using, what you want it to do, and show me what you typed. I'll show you exactly where it breaks and how to rebuild it.
"Write me a marketing email"
Generic 500-word blob. Wrong tone. No CTA. Misses the point entirely.
"You are a senior email copywriter for B2B SaaS. Write a 150-word re-engagement email for churned users. Tone: casual but urgent. Include one clear CTA button text. Format: subject line, preview text, body, CTA."
Precise. On-brand. Structured. A far stronger starting point than the blob.
// techniques_i_teach
The Toolkit Inside the Chat
Chain-of-Thought
Force the AI to reason step-by-step instead of leaping to conclusions. Fewer hallucinations, better logic.
Few-Shot Examples
Show the AI 2-3 examples of what you want. Pattern matching is how these models actually learn in-context.
Role Assignment
"You are a senior tax accountant" activates different knowledge clusters than "help me with taxes."
Constraint Setting
Word limits, formats, forbidden topics, required sections. Constraints are creative fuel for AI.
Output Formatting
Tell it you want JSON, markdown tables, bullet points, or numbered steps. Structure in = structure out.
Per-Model Tuning
What works on ChatGPT doesn't always work on Claude or Midjourney. Tell me which one you're using and I'll adapt the advice to it.
// sample_inputs
Things You Can Type Right Now
These are real messages you can send me in the chat. Copy-paste one or bring your own.
"I'm using ChatGPT to write product descriptions for my Etsy shop and they all sound generic. Here's my current prompt: [paste]. What am I doing wrong?"
→ I'll dissect it line by line and rebuild it"I want Midjourney to generate a moody noir detective scene but I keep getting cartoon stuff. Help me write the image prompt."
→ I'll teach you Midjourney's parameter syntax and style anchors"Explain chain-of-thought prompting like I'm 15. Then show me three examples I can actually use for homework help."
→ Plain-language breakdown with copy-ready templates"I'm building a system prompt for a customer support bot using Claude. It needs to be friendly but never promise refunds. Draft one for me."
→ Full system prompt with guardrails and personality tuning"What's the difference between prompting GPT-4 vs Claude vs Gemini? Like, do I need different strategies for each?"
→ Model-by-model comparison with practical tips// example_session
Example Conversation — Demo
A fictional sample exchange showing how a session flows. What you get depends on your prompt and the model you use it with.
"My blog intros from ChatGPT feel robotic. Current prompt: 'write an intro for my post about houseplants.' What am I doing wrong?"
Three gaps: no role, no audience, no tone. Try: "You are a gardening columnist for a casual lifestyle blog. Write a 100-word intro for a post about low-light houseplants, aimed at beginners who have killed every plant they owned. Tone: warm, a little self-deprecating." Paste the result back and we'll iterate.
"That already reads better. Why did naming the audience matter?"
Because "beginners who have killed every plant" tells the model which vocabulary, assumptions, and jokes fit. An audience is a constraint — and constraints are what keep output specific.
Example conversation — demo. Not a real user transcript.
// frequently_asked
Questions Before You Chat
Yes, and it's less meta than it sounds. Think of it like having a writing coach who specializes in a very specific format: AI instructions. Different models respond to different structures, and most people have never been taught how to write effective prompts. That's what this chat teaches you.
Text models like ChatGPT, Claude, Gemini. Image generators like Midjourney, DALL-E, and Stable Diffusion. Just tell me which one you're using and I'll tailor the advice. The core techniques carry over, but each model — and each new model version — behaves a little differently, so treat model-specific tips as starting points to test, not guarantees.
Absolutely. Paste your prompt, tell me what tool you're using and what result you wanted, and I'll rewrite it. But I'll also explain why the rewrite is better so you learn the pattern — not just get a one-time fix.
That's exactly who benefits most. If you've been typing things into ChatGPT and wondering why the output is meh, we can start there. No jargon required. I'll adjust the depth to wherever you are — just tell me your experience level.
Just a chat — no dashboard, no apps to install, no accounts to connect. You type messages, I respond with analysis, rewrites, and teaching. It's a conversation with a specialized AI on AURVEK's chat interface. That's it.
// initialize_session
Bring Your Worst Prompt.
I Dare You.
Paste it in. Tell me what AI you're using and what you expected. Let's see where the signal gets lost.
Start Chatting →No payment to open the chat — usage consumes your AURVEK wallet balance. You'll be in the chat in seconds after registering.