Prompt Engineering Patterns #

Working with LLMs has taught me that prompt engineering isn’t just about getting the right output — it’s about designing conversations that bring out the best thinking from both human and AI.

Core Patterns I’ve Discovered #

Chain of Thought — Breaking complex problems into steps
“Let’s think through this step by step…”

Role Playing — Giving the AI a specific persona or expertise
“You are a senior Tamil linguist with expertise in classical literature…”

Few-Shot Examples — Showing the pattern you want
“Here are 3 examples of the format I need…”

Constraint Setting — Clear boundaries and expectations
“Respond in exactly 100 words, using only technical terminology…”

What’s Emerging #

The most interesting development isn’t about prompts themselves — it’s about developing a shared language between human intuition and machine capability. We’re learning to think alongside AI, not just instruct it.

This requires a new kind of literacy: understanding not just what to ask, but how AI “thinks” about problems, what it struggles with, and where human insight remains irreplaceable.

Status: Still collecting patterns and testing approaches. This note will grow as I discover more.