Prompt Engineering: Getting AI to Understand What You Actually Need

"Prompt engineering" sounds technical, but it's really just the skill of asking questions in a way that gets you what you actually need. In the context of reentry and second chances, this is crucial—because how you phrase your situation to AI determines whether the output is generic or genuinely useful for your specific circumstances.

Think of it like the difference between asking a friend "Do you have advice?" versus "I'm dealing with X situation where Y happened, and I need help with Z." The second question gets a better answer because it's specific.

The Three Elements of a Strong Reentry Prompt

1. Context: Give AI the relevant background. Not your whole life story, but the framework it needs. "I have a felony conviction from 2018. I've completed anger management training and a vocational certification. I'm applying for warehouse supervisor roles."

2. Goal: Be explicit about what you want. "Help me write a background explanation letter that acknowledges my past honestly but emphasizes what I've learned and changed." This is clearer than "Help me write something about my background."

3. Constraints: Tell AI what matters to your situation. "The tone should be professional but not defensive." Or "This needs to be appropriate for a small, tight-knit team where trust matters." These constraints help AI tailor the output.

Why Specificity Changes Everything

When you're vague, AI generates generic output. When you're specific, you get personalized guidance. The difference: "Write an interview answer about my past" versus "I'm interviewing for a retail management role at a company that hires people with records. The hiring manager will likely ask about my conviction. Write a 60-second answer that takes responsibility, explains what led to it, and shows what I've learned and changed."

The second prompt gives AI the situation, the constraint (60 seconds), the context (what the hiring manager cares about), and the tone (take responsibility, but show growth). That produces something actually useful.

Common Weak Prompts (and How to Fix Them)

Weak: "Help me with my job application." | Strong: "I'm applying for entry-level assembly work. My last job was 4 years ago; I was incarcerated for 3 years since then. I've done a welding apprenticeship while inside. Help me write a cover letter that explains the gap naturally while highlighting my new skills."

Weak: "Write me a reference." | Strong: "Generate a character reference template for my former mentor to fill in. We worked together at a community center for 18 months on youth mentoring. The template should emphasize reliability, problem-solving, and integrity."

Iterating: The Real Power

One prompt is rarely perfect. The real skill of prompt engineering is iteration. AI gives you a draft. You read it and think, "That's close but sounds too formal" or "This doesn't mention [important thing]." So you refine: "That's helpful, but can you make it less formal and more direct? Also, add something about how I've handled setbacks."

This back-and-forth is how you turn generic AI output into something that's actually yours—your voice, your story, your strategy.

Try this: Take a situation you're facing (a background explanation, interview prep, reference guidance). Write your prompt three ways: vague, better, and strongest. Then use the strongest version with Claude. Notice how the output gets progressively more useful with specificity.