Every great AI prompt has three parts: Instructions (what to do), Context (background the AI needs), and Constraints (format, length, tone, what NOT to do). This isn't a theory — it's a repeatable structure that turns vague requests into specific, useful outputs every single time.
What Is the ICC Framework?
Instructions are the core task. Not "help me with marketing" but "write a 3-email nurture sequence for trial users who haven't converted after 7 days." The more specific the instruction, the less the AI has to guess.
Context is everything the AI can't infer. Your audience, your industry, what you've already tried, why you need this, what the output will be used for. The AI is smart but it can't read your mind — context fills the gap between what you typed and what you meant.
Constraints are guardrails. Word count, tone, format, what to avoid. "Under 200 words, professional but warm, no jargon, include a CTA." Without constraints, the AI defaults to verbose, generic, safe output.
A bad prompt gives the AI 100 possible interpretations. ICC narrows it to one. The time investment is 30 seconds. The quality improvement is dramatic.
How Does ICC Look in Practice?
Without ICC: "Write a blog post about remote work."
With ICC:
The second prompt takes 30 extra seconds to write. The output takes zero additional editing. That's the trade-off ICC makes for you.
When Should You Skip ICC?
Simple factual questions don't need it. "What's the capital of France?" doesn't benefit from constraints. Use ICC when you need the AI to create something — a document, analysis, email, code, plan, or recommendation. The more creative or subjective the task, the more ICC helps.
Save your best ICC prompts as templates. Next time you need a similar output, swap the context but keep the instructions and constraints. You'll build a personal library of prompts that produce consistently excellent results.
Try it now: Take the last prompt you typed into ChatGPT. Rewrite it with explicit Instructions, Context, and Constraints. Compare the outputs. The difference speaks for itself.