Let's Start Here |
You know that feeling when you ask AI for something organized and get back a wall of text? |
You're frustrated because you know what you want. The problem isn't your prompt. |
The problem is AI doesn't know your format. |
Until now. |
Here's what's about to happen: Over the next few minutes, you're going to learn a simple way to tell AI exactly what you want and how you want it. By the end of this email, you'll understand JSON. Not as a code. But as a way to communicate with AI effectively. |
|
|
📚 What You're Learning Today |
You're learning the building blocks of JSON. |
What those brackets are. How to read them. |
How to build them. |
By tomorrow, you'll be writing actual prompts. But today? We're laying the foundation so nothing feels scary. |
|
🎬 Here's A Real Situation |
Imagine you run a mid-sized SaaS company. Every day, your support team gets tickets. Each ticket is a customer explaining a problem. |
Right now, someone manually reads each one and fills out a spreadsheet: |
|
That's probably 30 minutes a day of pure data entry. Just organizing what customers already told you. |
Now imagine AI did that for you. |
Same information, perfectly organized, ready to go. |
That's what JSON prompting does. It tells AI: "Extract this, put it here, organize it like this." |
Real talk: I used to spend hours every week asking AI for information, then reformatting it into what I actually needed. One day I realized I was spending more time fixing AI's output than getting work done. So I learned JSON. Three months later, my AI output was already organized. No reformatting. Just usable work. |
|
|
|
💡 Here's The Core Idea |
JSON is a way to organize information so there's no confusion. Think of it like filling out a form at a doctor's office: |
Name: [your name goes here] Date of birth: [your date goes here] Reason for visit: [your reason goes here]
|
You don't write an essay in the "Name" field. |
You just write your name. |
Because the form told you what goes where. |
JSON is the same idea. You create labeled boxes. |
Each box has a purpose. |
AI fills them in correctly because the structure is clear. |
That's literally it. |
🔍 The Three Things JSON Uses |
JSON only needs three symbols. You already know what they are. |
You've seen them a thousand times: |
{ } Curly brackets = organized information container |
: Colon = label to value separator |
[ ] Square brackets = a list of items |
That's your entire toolkit. |
Three symbols, everything else is just how you combine them. |
|
🏗️ Let's See It In Action |
This is a simple JSON structure. Don't overthink it. Just read it: |
{
"customer_name": "Sarah Chen",
"email": "sarah@company.com",
"issue_type": "billing",
"description": "Charged twice for subscription",
"priority": "high"
}
|
See what's happening there? |
|
Now look at what it looks like when you add a list: |
{
"customer_name": "Sarah Chen",
"email": "sarah@company.com",
"issue_type": "billing",
"steps_to_fix": [ "Check billing page",
"Review payment history",
"Request refund",
"Update card on file" ],
"status": "in-progress"
}
|
The square brackets [ ] hold a list. Each item is separate. Still organized. Still clear. |
Key insight: You're not writing code. You're creating a template. A form. A structure that says "put information here, that information there." AI reads this structure and fills it in perfectly because it understands exactly what you want. |
|
|
📝 Understanding Placeholders |
When you write a JSON structure for AI, you don't know the values yet. That's why you use placeholders. These are instructions inside brackets that tell AI what goes there: |
❌ Vague Placeholder (AI might get confused): |
"description": "[description]"
|
AI might give you 1 sentence. Or 5. Or a paragraph. You don't know. |
✅ Clear Placeholder (AI knows exactly what to do): |
"description": "[2-3 sentences explaining the specific problem]"
|
AI knows: 2-3 sentences, focus on the problem. No guessing. |
The difference? Specificity. The more specific your placeholder, the better your result. |
|
🛠️ Your First JSON Template |
Here's a complete structure for capturing customer support tickets. This is something you could actually use: |
{
"customer_name": "[full name]",
"email": "[email address]",
"account_status": "[free / paid / enterprise]",
"issue_category": "[billing / technical / feature request / other]", "issue_description": "[2-3 sentences describing what happened]", "urgency_level": "[low / medium / high - explain why]", "requested_outcome": "[what the customer wants resolved]", "next_steps": [
"[first action required]",
"[second action if needed]" ]
}
|
Every field tells AI what type of information goes there. And when AI fills this in, you get perfectly organized output. |
Quick rule: Label names should be simple (no spaces, lowercase, with underscores if you need them). Like "customer_name" instead of "Customer Name". This keeps everything clean. |
|
|
❌ vs ✅ Why This Matters |
❌ Without JSON |
You ask AI: "Summarize this support ticket" |
You get: A paragraph. Information mixed together. You spend 10 minutes pulling out details and entering them into your system. |
✅ With JSON |
You ask AI: "Extract ticket info using this structure: { ... }" |
You get: Organized output. Fields already separated. Copy-paste ready. 30 seconds. |
That's 30 minutes saved per day. For your whole team. |
|
🚀 Your Challenge Right Now |
Here's what I want you to do: Think about something in your work where you organize information regularly. Could be: |
Customer feedback you need to categorize Meeting notes you need to summarize Project updates you track Job applications you review Anything where you're pulling out and organizing details
|
Now create a simple JSON structure for it. |
Just the labels and placeholders. |
No actual values yet. |
Use this as your template: |
{
"field_1": "[what information goes here?]",
"field_2": "[what information goes here?]",
"field_3": "[what information goes here?]",
"field_4": "[what information goes here?]"
}
|
That's it. 4 fields. Clear placeholders. |
You've got your first JSON structure. |
🎯 Remember This |
JSON isn't scary because it isn't complicated. It's literally labeled boxes. If you can name things and describe what goes in them, you can do this. |
You're not learning to code. You're learning to communicate what you want clearly. |
|
👀 What's Coming Tomorrow |
Tomorrow: You take what you learned today and build your first actual working prompt. The kind you paste into ChatGPT or Claude and get perfectly organized output. You'll see how theory becomes real. |
That's when the time-saving actually starts. |
✍️ Your Turn — I Want To Hear From You |
Pick whichever resonates: |
Option 1 — Show Me Your Structure: Reply with a JSON template you created for your work. I'll give you feedback and make sure it's solid for tomorrow's lesson. |
Option 2 — Ask Me Your Questions: Anything didn't click? Any part of the JSON feel unclear? Reply and ask. I'll explain it simpler. |
Option 3 — Tell Me Your Work: What's your industry? What do you do? Tell me and I'll make tomorrow's example speak directly to your situation. |
Seriously—reply. I read every message, and this conversation helps me make the rest of the week even better for you. |
|
You just learned JSON. Not as a programmer would. As someone who needs to use AI effectively would. |
You understand the structure. You know what the symbols mean. You know how to build one. |
That foundation takes you from "I'm confused by AI formatting" to "I can make AI output exactly what I need" in just a few days. |
Tomorrow you start building. But today? Today you understand. |
— Prompt Guy |
P.S. The best part about understanding JSON is it's not something you forget. Once you see "oh, it's just labeled boxes," you realize how simple it was. That moment is coming for you tomorrow when you write your first actual prompt. |
