PROMPTING
PROMPTING
PROMPTING
For this project, my goal was simple:
I wanted the AI to generate a structured, SEO-friendly JSON sitemap directly from a business idea. Not just suggestions or explanations, but actual usable data that could plug straight into my workflow.
For this project, my goal was simple:
I wanted the AI to generate a structured, SEO-friendly JSON sitemap directly from a business idea. Not just suggestions or explanations, but actual usable data that could plug straight into my workflow.
On the left side, you can see the result after I carefully engineered the prompt.
I told the AI to act as a web strategist, SEO specialist, and information architect. I clearly defined its role, gave it a structured task, and specified the output format as JSON.
Because of that, the AI gave me exactly what I needed: a full website structure, clearly organized in JSON, with SEO-focused titles and descriptions already included.
On the left side, you can see the result after I carefully engineered the prompt.
I told the AI to act as a web strategist, SEO specialist, and information architect. I clearly defined its role, gave it a structured task, and specified the output format as JSON.
Because of that, the AI gave me exactly what I needed: a full website structure, clearly organized in JSON, with SEO-focused titles and descriptions already included.
But on the right side, when I used a basic prompt without clear instructions, the AI defaulted to what it thinks is helpful: general advice, some website strategy, and steps I should follow. It didn’t give me the structured output I needed.
So, without prompt engineering, it just acted like a consultant rather than a tool I could build with.
But on the right side, when I used a basic prompt without clear instructions, the AI defaulted to what it thinks is helpful: general advice, some website strategy, and steps I should follow. It didn’t give me the structured output I needed.
So, without prompt engineering, it just acted like a consultant rather than a tool I could build with.
That’s why prompt engineering matters. It’s not just about asking the AI for help—it’s about controlling how the AI thinks, so you can use it as a tool, not just a chatbot.
That’s why prompt engineering matters. It’s not just about asking the AI for help—it’s about controlling how the AI thinks, so you can use it as a tool, not just a chatbot.
Tell the AI who it is.
Tell the AI who it is.
Be specific about the task.
Be specific about the task.
Clearly explain what you expect as output.
Set up boundaries for the AI.
Work
AI
AI
GET IN TOUCH
On the left side, you can see the result after I carefully engineered the prompt.
I told the AI to act as a web strategist, SEO specialist, and information architect. I clearly defined its role, gave it a structured task, and specified the output format as JSON.
Because of that, the AI gave me exactly what I needed: a full website structure, clearly organized in JSON, with SEO-focused titles and descriptions already included.
But on the right side, when I used a basic prompt without clear instructions, the AI defaulted to what it thinks is helpful: general advice, some website strategy, and steps I should follow. It didn’t give me the structured output I needed.
So, without prompt engineering, it just acted like a consultant rather than a tool I could build with.
That’s why prompt engineering matters. It’s not just about asking the AI for help—it’s about controlling how the AI thinks, so you can use it as a tool, not just a chatbot.
For this project, my goal was simple:
I wanted the AI to generate a structured, SEO-friendly JSON sitemap directly from a business idea. Not just suggestions or explanations, but actual usable data that could plug straight into my workflow.
Set up boundaries for the AI.
Work
Home
AI
Work
About
GET IN TOUCH
Clearly explain what you expect as output.
Set up boundaries for the AI.
Be specific about the task.
Tell the AI who it is.
Clearly explain what you expect as output.