Session 6: Prompt Engineering — Talking to AI Like a Pro

The difference between a mediocre AI answer and a mind-blowing one is rarely the model — it's the prompt.

What Is Prompt Engineering?

Prompt engineering is the art and science of crafting instructions that get the best possible output from an AI model. It's less "engineering" and more "clear communication" — think of it like being a great manager who gives crystal-clear directions.

Why it matters: The same LLM given a vague prompt versus a well-crafted prompt will produce dramatically different results. The model hasn't changed — only your instructions did.

Analogy: Prompt engineering is like giving directions. "Go that way" will get someone lost. "Turn left on Oak Street, drive 3 blocks, turn right at the blue building" gets them exactly where they need to be.

The 5 Building Blocks of a Great Prompt

1. Role — Tell AI who to be. ("You are an experienced copywriter specializing in SaaS landing pages...")
2. Context — Provide the background information the AI needs to give a relevant answer.
3. Task — State clearly and specifically what you want. The more precise, the better.
4. Constraints — Set boundaries: length, format, tone, what to include or exclude.
5. Examples — Show what good output looks like. Nothing communicates expectations like a concrete example.

Prompting Techniques

Zero-Shot Prompting

Just ask directly with no examples provided. Works well for simple, well-defined tasks where the AI already understands what you need.

Example: "Summarize this article in 3 bullet points."

Few-Shot Prompting

Provide 2–3 examples of the input-output pattern you want, then give the AI a new input to complete.

Example:

• Casual: "Hey, gonna be late" → Formal: "I wanted to inform you that I will be arriving later than expected."
• Casual: "Can't make it tomorrow" → Formal: "I regret to inform you that I will be unable to attend tomorrow."
• Casual: "Sounds good, let's do it" → Formal: "That sounds agreeable. Let's proceed."
• Casual: "Need this done ASAP" → Formal: (AI completes the pattern)

Chain-of-Thought Prompting

Ask the AI to think step by step before giving its final answer. This dramatically improves accuracy on reasoning tasks, math problems, and complex analysis.

Key phrase: "Think through this step by step before giving your final answer."

By forcing the model to show its work, you reduce errors and get more reliable outputs — especially for multi-step problems.

Role Prompting

Assign the AI a specific identity, profession, or perspective. This changes its tone, vocabulary, depth, and overall approach.

Example: "You are a senior data scientist explaining to a marketing team why the latest A/B test results are statistically significant."

Common Prompt Mistakes

Advanced Tips

• Iterate, don't one-shot. Treat AI like a collaborative drafting partner, not a vending machine.
• Ask AI to critique its own work: "Review your response. What could be stronger?"
• Negative prompting: "Don't use jargon," "Avoid clichés," "Do not include a greeting."
• Specify output format: "Respond as JSON," "Use markdown with H2 headings," "Return a numbered list."
• Multi-turn refinement: Use follow-ups to revise, redirect focus, expand sections, or trim length.

Real-Life Examples

• Marketing manager: Uses role prompting to generate email subject lines for a specific audience segment — "You are a conversion-focused email marketer writing to busy executives."
• Student: Uses chain-of-thought for physics problems — "Solve this step by step, showing each formula and substitution."
• Recruiter: Uses few-shot prompting with example job descriptions to generate consistent new postings in the same style.
• Content creator: Uses iterative refinement — "Write a draft → Make it funnier → Stronger hook → Trim by 20%."
• Developer: Uses constraint-heavy prompts — "Write a Python function that validates email addresses using regex. Include type hints, a docstring, handle edge cases, and use no external libraries."