Free Guide · No Email Required

The Deep Think Prompt Guide

10 advanced techniques to get dramatically better results from any AI — the mechanics behind real prompt engineering, not just more prompts.

Most people use AI like a search engine: type a question, take the first answer, move on. That works fine for simple tasks. But there's a second tier of prompting most beginners never discover — where you stop asking the AI for answers and start engineering conversations.

This is what “Think” really means in the Click. Think. Create. framework. Not just what you ask — but how the asking itself is structured.

Each technique below teaches a mechanic. Master the mechanic, and you can apply it to any topic, any tool, any task — far beyond the specific examples here.

Technique 01

The Interrogation Flip

The Mechanic

Instead of giving the AI all your information upfront, make it ask you the questions first.

Why It Works

Most people write bloated, over-explained prompts because they're guessing what details matter. Flip it — let the AI tell you what it needs. You'll often get asked about details you hadn't even considered relevant.

◆ Copy this prompt
You are a senior strategist helping me think through [DECISION/PROBLEM]. Do not give me advice yet. Instead, ask me one question at a time — the single most important thing you need to know to give me useful guidance. Wait for my answer before asking the next question. After 4-5 questions, synthesize everything into a clear recommendation.
Technique 02

The Constraint Stack

The Mechanic

Layer multiple specific constraints into a single prompt — role, format, length, tone, and exclusions — all at once, rather than one vague instruction.

Why It Works

A single unconstrained prompt gives the AI total freedom, which usually produces generic output. Constraints aren't limitations — they're the steering wheel.

◆ Copy this prompt
Act as [SPECIFIC EXPERT ROLE]. Write [OUTPUT] about [TOPIC] for [SPECIFIC AUDIENCE]. Length: under 200 words. Tone: direct, no hedging language. Do not use the words “leverage,” “robust,” or “seamless.” Include exactly one concrete example. End with a single actionable next step, not a summary.
Technique 03

The Devil's Advocate Loop

The Mechanic

Ask the AI to argue against your own position — then ask it to argue for the opposite, then synthesize.

Why It Works

AI tools default to agreeing with your framing. This forces genuine adversarial thinking, which surfaces blind spots a single straightforward question never will.

◆ Copy this prompt
I believe [YOUR POSITION/PLAN]. Argue against this as convincingly as possible — find every weakness, risk, and flawed assumption. Don't soften it. When you're done, switch sides completely and argue for my original position just as forcefully. Then give me a one-paragraph synthesis of what both arguments reveal.
Technique 04

The Persona Stack

The Mechanic

Run the same question through multiple distinct expert personas in one sequence, then compare their answers side by side.

Why It Works

A single “expert” framing only gives you one lens. Stacking personas surfaces how different fields would actually approach the same problem differently — which is closer to how real decisions get made.

◆ Copy this prompt
I want three perspectives on [PROBLEM/DECISION]. First, answer as a [ROLE 1] would. Then, completely reset and answer as a [ROLE 2] would. Then, reset again and answer as a [ROLE 3] would. Keep each answer under 100 words. After all three, identify where they agree and where they sharply disagree.
Technique 05

The Compression Test

The Mechanic

Ask for a long answer first, then force progressively shorter versions of the same answer.

Why It Works

Each compression pass strips away padding and forces the real core idea to surface. The 1-sentence version is often the most useful thing you get out of the whole exchange.

◆ Copy this prompt
Explain [TOPIC] in a thorough paragraph. Now compress that into 3 bullet points. Now compress those 3 bullets into a single sentence. Now compress that sentence into 5 words or fewer.
Technique 06

The Pre-Mortem

The Mechanic

Ask the AI to imagine your plan has already failed, then work backward to explain why.

Why It Works

People are much better at predicting causes of failure when failure is framed as already having happened, rather than being asked to assess risk in the abstract.

◆ Copy this prompt
Imagine it's one year from now and [YOUR PLAN/PROJECT] has completely failed. Write a short retrospective explaining what went wrong, as if you're looking back with full hindsight. Be specific about the failure points, not generic. Then list the top 3 things I could do now to prevent each one.
Technique 07

The Socratic Refusal

The Mechanic

Explicitly forbid the AI from giving you the answer — instruct it to only ask guiding questions until you arrive at the answer yourself.

Why It Works

This is the single best technique for actually learning something rather than just outsourcing the thinking. You retain what you discover yourself far better than what you're told.

◆ Copy this prompt
I want to understand [CONCEPT/PROBLEM] deeply, not just get an answer. Do not explain it to me directly. Instead, ask me a series of guiding questions, one at a time, that lead me to figure it out myself. Only confirm whether my reasoning is on track — don't give away the conclusion.
Technique 08

The Constraint Removal

The Mechanic

Ask for the realistic answer first, then explicitly ask what the answer would be with constraints removed — budget, time, convention, fear of failure.

Why It Works

People (and AI, mirroring them) self-censor toward “reasonable” ideas by default. Explicitly removing constraints in a second pass surfaces the ambitious option that was filtered out the first time.

◆ Copy this prompt
Give me a realistic plan for [GOAL] given normal constraints. Now ignore budget, time, and conventional wisdom entirely — what would the most ambitious version of this look like? Compare the two and tell me which elements of the ambitious version are actually achievable if I stretched.
Technique 09

The Translation Chain

The Mechanic

Force the same idea to be re-explained for completely different audiences in sequence — expert, beginner, child, skeptic.

Why It Works

If you can't simplify an idea, you don't fully understand it yet. This technique is as much a comprehension test for you as it is a communication tool.

◆ Copy this prompt
Explain [CONCEPT] four times in a row, each for a different audience: 1) a fellow expert in the field, 2) a complete beginner, 3) a curious 10-year-old, 4) a skeptic who thinks the whole idea is overrated. Keep each version under 60 words.
Technique 10

The Self-Critique Pass

The Mechanic

After getting an answer, ask the AI to critique its own output as if reviewing someone else's work.

Why It Works

AI tools rarely flag weaknesses in their own first response unprompted. Asking for a self-critique as a separate, distinct step produces a noticeably more honest evaluation than asking it to “double check” in the same breath.

◆ Copy this prompt
Here is your previous response: [PASTE RESPONSE]. Now critique it as if you were a tough, skeptical editor reviewing someone else's work. What's weak, vague, or unsupported? What's missing? Then rewrite the response addressing every issue you just raised.

These 10 techniques are just the start.

The Practical AI Guide for Beginners goes further — 200 ready-to-use prompts across work, writing, learning, creativity, real estate, careers, health, and side hustles, plus a 30-day action plan and a complete 12-tool AI toolkit.

12Tools explained
30Day action plan
200Ready-to-use prompts
Get the book on Amazon →