Prompt Intelligence Series · 2026

Prompt
Intelligence

The Insider Playbook for ChatGPT, Claude, Gemini and Perplexity

Hidden rules. Model-specific techniques. What each model variant actually does. Prompt security. Twelve copy-paste frameworks. Everything the documentation does not tell you.

ChatGPT Claude Gemini Perplexity
Prashant Shinde Founder and CEO, Agile Visa · ICAgile Accredited Provider

Introduction

Why 95% of People Get Mediocre Results from AI

The gap between a novice and an expert AI user is not about magic prompts. It is about setup.

Here is what the AI companies do not advertise. Every large language model has a hidden architecture of priorities, constraints, and behavioural patterns that determine the quality of its output. Most users interact with the surface layer. This guide takes you underneath it.

When someone types a one line question into ChatGPT and gets a generic answer, they assume the AI is limited. When a power user types a precisely constructed prompt into the same model and gets a response that feels like it was written by the smartest person in the room, they understand something different. The model did not change. The interaction architecture did.

The Single Most Important Insight in This Guide

The difference between an expert AI user and an average one is not knowledge of magic prompts. It is the habit of spending 2 minutes setting up context before asking anything. An expert writes a 5 line context block at the start of every important conversation. A novice types one sentence and wonders why the output is mediocre.

The Universal Context Block

Before you ask any substantive question in any model, paste this block and fill it in:

Universal Context Block — Use Before Every Important Prompt
I am [your role] at [type of organisation]. My goal right now is [specific outcome]. My audience for this output is [who will read/use it]. The format I need is [document / bullet list / email / analysis / code]. The tone should be [formal / direct / conversational / technical]. Things to avoid: [anything the model should not do or assume]. Here is the task: [your actual request]
This single habit improves output quality more than any prompt technique in this guide.
01
Context Ceiling Rule
The quality of output can never exceed the quality of context you provide. The AI is not limited. Your setup is.
02
Priority Stack Architecture
Every model has a hidden hierarchy of what it optimises for first. Understanding this predicts why your prompts succeed or fail.
03
The Opening Ceiling
How you open a conversation sets the ceiling for everything that follows. A weak opening can never produce a strong response, no matter what you ask later.
04
Negative Constraints Work
Telling the model what NOT to do is as powerful as telling it what to do. Most people only use positive instructions.

Part One

The Hidden Rules
What Each Model Actually Prioritises

Before you write a single prompt, understand what the model is optimised to do.

Every model has a priority stack. This is the hierarchy of objectives it tries to satisfy simultaneously. Understanding this hierarchy explains why prompts succeed or fail in ways that feel arbitrary but are completely predictable once you know the pattern.

ChatGPT Claude Gemini Perplexity
Priority 1 Helpfulness Safety + accuracy Accuracy + grounding Citation quality
Priority 2 User satisfaction Nuanced reasoning Helpfulness Answer completeness
Priority 3 Task completion Honesty Google integration Source diversity
Weakness Can overclaim Can over-caveat Can over-cite Less creative depth
What This Means Practically

ChatGPT will give you a confident-sounding answer even when it should express uncertainty. Claude will add caveats even when you do not need them. Gemini will cite sources even when you just want a direct answer. Perplexity will surface great evidence but struggle with synthesis. Counter each weakness with a direct instruction.


Part Two

ChatGPT

The most widely used model. Also the most misused. Here is what the documentation does not tell you.

How ChatGPT Actually Processes Your Prompt

ChatGPT uses Reinforcement Learning from Human Feedback, which means it was trained to produce outputs that human raters preferred. This creates a specific pattern. It is wired to produce responses that feel satisfying and complete even when the honest answer is uncertainty. This is why it sometimes sounds more confident than it should be.

Hidden Behaviour — ChatGPT Confidence Bias

ChatGPT has a strong bias toward producing a full answer even when it should say it does not know. Counter this explicitly: tell it "If you are uncertain about any part of this, say so explicitly rather than filling the gap with an assumption."

The Custom Instructions Feature Most People Ignore

In ChatGPT's Settings menu there is a section called Custom Instructions. Two boxes. These instructions run as a silent system prompt on every single conversation. Most users have never opened this menu. The people who fill it in get dramatically better outputs every time because the model already knows who they are before they say a word.

ChatGPT Custom Instructions — Box 1 (Who You Are)
I am a [role] specialising in [domain]. I work with [type of clients or organisation]. I have [X] years of experience in [field]. I use AI daily for [specific use cases]. I prefer direct, concise answers. I do not need lengthy preamble. I can handle technical depth. Do not simplify unless I ask.
ChatGPT Custom Instructions — Box 2 (How to Respond)
Always lead with the most important point. Use plain language unless technical precision is required. Format in prose unless I specifically ask for bullet points. If you are uncertain, say so. Do not fill gaps with confident-sounding assumptions. When I ask for analysis, give me your actual view, not a balanced both-sides summary unless I ask. Never add a conclusion paragraph that repeats what was already said.
These two boxes take 20 minutes to fill in once and work silently on every conversation forever.

Strong vs Weak Role Assignment

Weak — Produces Generic Output
Act as a marketing expert and help me write a campaign strategy.
Strong — Produces Specific, High-Quality Output
You are the CMO of a professional training company with 5,000 alumni and a $0 paid acquisition budget. You have 6 weeks to generate revenue from an existing email list of warm contacts. You have used paid ads before and they did not work. You are not interested in theoretical frameworks. You need a 30-day action plan with specific daily tasks that a solo founder can execute alone. Give me that plan.
The constraints and context force the model out of its default generic answer.
Technique — Force Contrarian Thinking

After any analysis, add: "Now argue the opposite. What is the strongest case against everything you just said?" This forces ChatGPT out of its confirmation bias and gives you a complete picture of any situation.


Part Three

Claude

Claude thinks differently from other models. Understanding how it reasons is the key to unlocking its full capability.

Claude is built by Anthropic with a strong emphasis on reasoning quality and intellectual honesty. Unlike models optimised primarily for user satisfaction, Claude is trained to express genuine uncertainty, push back on incorrect premises, and think through problems rather than pattern-match to a likely-sounding answer. This makes it the most reliable model for tasks where accuracy matters more than speed.

Extended Thinking Mode

Claude has a mode called extended thinking where it reasons through a problem in a hidden scratchpad before responding. The response you see is the output of that reasoning process, not a first-pass pattern match. When the toggle is on, Claude visibly shows its reasoning in a collapsible section before the answer. Two things make extended thinking work better. First, the prompt needs to be genuinely complex. Simple questions get a short think regardless. Second, including explicit reasoning constraints in your prompt makes the thinking deeper.

Trigger Deeper Reasoning in Claude
Before you answer, think through the following: 1. What are the key assumptions I might be making that could be wrong? 2. What are the 2 or 3 most important tradeoffs here? 3. What would the best expert in this field say that a generalist would miss? Then give me your answer with your reasoning visible.
This forces Claude into rigorous thinking mode even without the extended thinking toggle enabled.

The Projects Feature — Your Permanent Context Layer

Claude Projects is the most underused feature in the entire AI landscape. Create a project, upload documents, write a system prompt, and every conversation in that project starts with full context. Claude knows your business, preferences, constraints, and goals before you say a word. For founders, consultants, and professionals with complex ongoing work, this feature alone is worth the Claude Pro subscription.

What to Put in Your Claude Project Document

Your role and background. Your company and what it does. Your primary goals right now. Your communication preferences. What you want Claude to always do and never do. Key facts about your situation that would take 10 minutes to re-explain every session. Update this monthly. It is your permanent AI memory.

The Devil's Advocate Prompt — Use Only With Claude
I am about to make [decision / investment / hire / strategic move]. Here is my current thinking: [your reasoning] I want you to be a rigorous devil's advocate. Do not be polite. Find every flaw in my thinking. Tell me what additional information I need before making this decision. Finally, give me your honest view on whether this is the right move.
Claude handles this better than any other model because it is designed not to tell you what you want to hear.

Part Four

Gemini

Gemini is not trying to be ChatGPT. It is built for a different job. Once you understand that job, it becomes remarkably powerful.

Gemini is Google's model and it is architected around Google's core advantage which is access to real-time information and deep integration with the Google product ecosystem. Where ChatGPT and Claude work primarily from their training data, Gemini can access current web information and connect to your Google Workspace in ways the other models cannot.

The Multimodal Secret Most People Miss

Gemini 1.5 Pro can process an entire PDF, video, audio file, or image alongside your text prompt in a single request. Upload a 50-slide competitor deck and ask "What is their positioning strategy and where are the gaps?" Or upload a meeting recording and ask "What were the three most important decisions made and who is responsible for each?" No other model handles this combination as well.

Google Search Grounding

Gemini has a feature called Google Search grounding which connects its responses to live search results. When this is on, Gemini's answers are anchored to current information rather than training data. This is critical for market research, competitive analysis, recent news, or anything time-sensitive. Without grounding, Gemini behaves like any other model with a knowledge cutoff. With grounding, it behaves like a research analyst with live internet access.

Gemini Deep Research Prompt (With Grounding On)
I am doing competitive research on [topic / company / market]. Please search for and synthesise: - Their recent announcements or news in the last 6 months - How their positioning has changed compared to 12 months ago - What customers and reviewers are saying about them right now - Any pricing or product changes recently Present this as a structured briefing, not a list of links.

Part Five

Perplexity

The only model in this guide built primarily around search. Most people use it like a search engine. That leaves most of its capability on the table.

Perplexity sits at the intersection of a search engine and a language model. Every answer it produces is grounded in live web sources which it cites inline. This means you can verify every claim it makes in real time. For professional work where accuracy and attribution matter, this is a critical advantage.

Pro Search Mode and Focus Modes

In Pro search mode, Perplexity asks clarifying questions before answering, runs multiple searches across different angles, and synthesises a more comprehensive answer. Always use Pro search for anything that matters. The Focus modes restrict search to specific sources: Academic searches Semantic Scholar and academic databases. Reddit Focus searches Reddit discussions which is surprisingly useful for understanding what real users actually think about products rather than what companies say about them. News Focus restricts to recent sources only.

Advanced Technique — The Multi-Angle Research Method

Search the same topic three times in Perplexity with three different Focus modes. Academic mode gives you what researchers say. Reddit mode gives you what practitioners and users say. News mode gives you what is happening right now. The gaps and contradictions between these three answers are often where the most interesting insights live.

Perplexity Structured Research Prompt
Using academic and professional sources where possible, research [your topic]. I specifically need: 1. A factual overview supported by recent sources 2. The main schools of thought or competing perspectives 3. The most credible recent data or statistics available 4. Names of key researchers, practitioners, or organisations I should know For each claim, cite the source so I can verify.
Perplexity is the only model where you should routinely check the citations.

The Chapter That Changes How You Choose

Model Guide
What Each Version Actually Does

Every AI company offers multiple model versions. Knowing which one to use for which task is as important as knowing how to prompt.

When you open ChatGPT or Claude, you are not just choosing a product. You are choosing from a family of models, each built for a different speed and intelligence tradeoff. Picking the wrong model for a task is like bringing a sledgehammer to a task that needs a scalpel. Understanding this changes how you work.

ChatGPT
by OpenAI
Everyday
GPT-4o mini
The fast, lightweight model. Handles most everyday tasks, quick Q&A, drafting, and simple analysis without burning through your usage limits.
Best for
Quick drafts Simple Q&A Summarising
Flagship
GPT-4o
The main model most people use. Strong across text, code, and images. Handles complex tasks, long documents, and multi-step reasoning well.
Best for
Complex writing Coding Strategy Image analysis
Reasoning
o3 / o4-mini
OpenAI's reasoning-focused models. They think before they answer, which takes longer but produces significantly better results on problems that require logic, maths, code debugging, or multi-step planning.
Best for
Logic problems Debugging Research synthesis Complex analysis
Claude
by Anthropic
Fast
Claude Haiku 3.5
Anthropic's fastest and lightest model. Built for speed at scale. When you need quick responses, large volume tasks, or API integrations where latency matters, Haiku is the right choice.
Best for
High volume tasks Quick summaries Chatbot integrations
Balanced
Claude Sonnet 4.6
The model most users interact with in Claude.ai. It balances deep reasoning with practical speed. Strong at writing, analysis, coding, and long document work. This is what powers most Claude Pro conversations.
Best for
Writing Analysis Coding Long documents
Most Capable
Claude Opus 4.6
The most intelligent and capable Claude model available. Built for the tasks that genuinely require the deepest reasoning. When nuance, precision, and intellectual rigour matter more than speed, Opus is the right choice.
Best for
Complex reasoning Research Strategic analysis Devil's advocate
Gemini
by Google
Fast
Gemini 2.0 Flash
Google's speed-optimised model. Designed for tasks where you need a quick answer grounded in real-time information. It still has access to Google Search, which means even the fast version stays current.
Best for
Quick searches Current news Fast drafts
Flagship
Gemini 2.5 Pro
Google's most powerful general model. Handles a 1 million token context window which means it can process entire codebases, long research documents, or hours of video in one prompt. The strongest multimodal model in this guide.
Best for
Multimodal tasks Video analysis Massive documents Workspace integration
Deep Research
Gemini Deep Research
A specialised mode rather than a separate model. When you activate Deep Research in Gemini Advanced, the model spends time browsing the web across multiple sources before synthesising a comprehensive, cited report. Think of it as an AI research analyst that does the legwork for you.
Best for
Market research Competitive analysis Topic deep dives
Perplexity
by Perplexity AI
Standard
Standard Search
A quick search with citations. Runs a web search, reads the top results, and gives you a synthesised answer with links. Good for fast fact-checking or getting a quick overview of any topic.
Best for
Fact checks Quick overviews Current events
Pro Search
Pro Search Mode
Asks clarifying questions first. Then runs multiple searches across different angles before synthesising. The quality jump over standard search is significant. For any research task that matters professionally, Pro Search is the minimum bar you should set.
Best for
Professional research Multi-angle topics Cited analysis
Focused
Focus Modes
Restrict search to specific source types. Academic mode pulls from Semantic Scholar and research databases. Reddit Focus surfaces community and practitioner discussion. News Focus locks to recent coverage only. YouTube Focus finds video content. Each gives you a completely different lens on the same topic.
Best for
Academic sources Practitioner views Breaking news Video research
The Model Selection Rule

Start with the standard model for each tool. Switch to the more powerful version when you hit a wall. For Claude this means starting with Sonnet and switching to Opus when the reasoning needs to go deeper. For ChatGPT it means using GPT-4o for most things and switching to o3 when you need genuine logic. For Gemini it means using Pro when the document is large or multimodal. Knowing when to upgrade your model is a skill in itself.


Part Six — The Chapter Most Guides Skip

Prompt Security

This is the chapter that could save your professional reputation.

Every time you paste text into a free AI tool, that text is potentially stored, reviewed, and in some configurations used to train future versions of the model. For professional tasks involving client information, financial data, strategic plans, or personnel matters, this is a serious risk that most organisations have not adequately addressed.

What You Paste and the Risk Level

Critical Risk
M&A or strategic plans
Personnel information
Unreleased financial data
High Risk
Client names with context
Financial figures
Legal documents
Medium Risk
Internal system details
Competitive strategies
Product roadmaps
Low Risk
General industry knowledge
Your own writing style
Public information

The Anonymisation Protocol

The most practical way to use AI safely with sensitive material is to strip all identifying information before pasting and replace it with generic placeholders. The AI does not need to know the client's name to help you structure a proposal.

Anonymisation Protocol — Use Before Pasting Sensitive Content
BEFORE PASTING — Replace all of the following: Client or company names → [CLIENT A], [CLIENT B] Specific financial figures → [REVENUE FIGURE], [COST FIGURE] Employee names → [PERSON A], [PERSON B] Product names if proprietary → [PRODUCT X] Location details if identifying → [REGION] AFTER AI RESPONSE: Manually reinsert the real details in your final document. This takes 3 minutes and eliminates 95% of your data leakage risk.

How to Opt Out of Training Data Usage

Platform Where to Go Toggle Off
ChatGPT Settings → Data Controls "Improve the model for everyone"
Claude Settings → Privacy Data usage for training toggle
Gemini Google Account → Data and Privacy → Web and App Activity Gemini Apps Activity
Perplexity Settings → Account Data collection preferences
The Enterprise Solution

ChatGPT Enterprise, Claude for Enterprise, and Gemini Workspace for Business all contractually guarantee that your data is not used for training. Zero exceptions. This is the single most important upgrade for any organisation handling client or financial data in AI tools. When you pitch AI upskilling to corporate clients, this is the data privacy answer they need before they say yes.


Part Seven

12 Master Templates

Copy-paste prompt frameworks for 12 professional scenarios. Fill in the brackets. Press enter.

01 of 12
The Strategic Decision Framework
Best with Claude
Copy Prompt
I need to make a decision about [decision topic]. Context: [2 to 3 sentences about the situation] Options I am considering: [option 1] / [option 2] / [option 3] Constraints: [time / budget / people / political] What I am optimising for: [primary goal] Analyse each option against the constraints and goal. Give me your clear recommendation with your reasoning. Tell me what the biggest risk of your recommended option is.
02 of 12
The Email Tone Calibration Prompt
Best with ChatGPT or Claude
Copy Prompt
Here is an email I need to send: [paste your draft] The recipient is: [their role and your relationship] The situation is: [context and stakes] I want the tone to be: [assertive / diplomatic / warm / direct / urgent] Things to preserve: [any specific phrases or commitments] Rewrite this email to hit exactly that tone without changing the substance.
03 of 12
The Research Synthesis Prompt
Best with Perplexity Pro
Copy Prompt
I am researching [topic] for [purpose]. Here is what I already know: [your existing knowledge] Here is what I specifically need to find out: [your questions] The format I need: [bullets / narrative / table / briefing doc] The audience for this: [who will read it] Synthesise what is known focusing on my specific questions. Highlight where there is genuine uncertainty or conflicting expert opinion. End with the 3 most important things I should know.
04 of 12
The Honest Feedback Extractor
Best with Claude
Copy Prompt
Here is a piece of work I have produced: [paste your work] The goal of this work was: [what it needed to achieve] The audience is: [who will receive it] Give me feedback in three parts: 1. What is working well and why it is effective 2. What is not working and specifically why 3. The single most important change I should make Do not be diplomatic. Give me honest professional feedback.
05 of 12
The Competitive Intelligence Brief
Best with Perplexity or Gemini
Copy Prompt
I need a competitive intelligence brief on [company / product / service]. My organisation is [brief description] competing in [market]. Specifically I want to understand: - Their current positioning and messaging - Their pricing model if available - Their apparent strategic direction - Where they appear strong - Where they appear vulnerable Format as a structured briefing I can share with my team.
06 of 12
The Meeting Preparation Prompt
Best with ChatGPT or Claude
Copy Prompt
I have a meeting with [person / company / role] in [timeframe]. The purpose is: [goal of the meeting] What I know about them: [background info] The outcome I want: [what success looks like] Prepare me for this meeting: 1. Three things I should know or research before going in 2. Five questions I should ask that a smart advisor would think of 3. The most likely objection or pushback I will face and how to handle it 4. One thing I should not say or do in this meeting
07 of 12
The Document Analysis Prompt
Best with Claude — 200K context window
Copy Prompt
Here is a document I need you to analyse: [paste document] 1. Give me a 3 sentence summary of the main point 2. Identify the 3 most important facts, claims, or commitments 3. Flag anything that is ambiguous, contradictory, or missing 4. Tell me what the author is NOT saying that I should probably know 5. Give me the 2 questions I should ask after reading this
08 of 12
The Persuasion Architecture Prompt
Draft with ChatGPT, refine with Claude
Copy Prompt
I need to persuade [audience] to [accept / approve / support] [your proposal]. Their current position is: [what they currently think] The main resistance I expect: [objections or concerns] What they care most about: [their priorities and values] The format is: [email / presentation / proposal / conversation] Write a persuasion structure that: - Meets them where they are rather than arguing against their position - Leads with what matters to them - Handles the resistance before they raise it - Ends with a clear and easy next step
09 of 12
The Crisis Communication Prompt
Best with Claude — it will not help you deceive
Copy Prompt
A situation has occurred: [describe the situation factually] Affected parties: [who is impacted] What happened: [facts only, no spin] What we are doing about it: [actions taken or planned] Audience for this communication: [customers / staff / board / public] Write a crisis communication that: - Acknowledges the situation directly without deflecting - States what is being done in concrete terms - Does not over-promise or under-deliver - Ends with a clear next communication commitment
10 of 12
The Learning Accelerator
Works well across all four models
Copy Prompt
I need to learn [subject or skill] in [timeframe]. My current knowledge level is: [beginner / some familiarity / intermediate] I learn best through: [examples / frameworks / stories / application] I will use this knowledge for: [specific application] Create a learning plan that: 1. Identifies the 20% of concepts that will give me 80% of capability 2. Gives me 3 things to read, watch, or do in the first 48 hours 3. Gives me a practical exercise I can do this week 4. Tells me the most common mistake beginners make and how to avoid it
11 of 12
The Pricing Strategy Prompt
Breadth with ChatGPT, depth with Claude
Copy Prompt
I am pricing [product / service / programme] for [target customer]. My costs are approximately: [rough cost structure] Competitors charge: [if known] My differentiation is: [why I am different or better] My customer's alternative is: [what they do if they do not buy from me] Help me think through: 1. What pricing model makes most sense and why 2. What the psychological anchoring should be 3. How to present the price so it feels like a clear decision 4. What a premium tier could look like to increase average order value
12 of 12
The LinkedIn Post Framework
Best with Claude — least AI-sounding output
Copy Prompt
I want to write a LinkedIn post about [topic / insight / experience]. My audience is: [professional audience] The insight I want to leave them with: [one clear takeaway] My personal angle on this: [what makes my perspective unique] Write a LinkedIn post that: - Opens with a line that makes a professional stop scrolling - Does not start with I and does not ask a question - Tells a real story or shares a real insight, not a generic observation - Never uses bullet points - Ends with a line that makes the reader think rather than a call to action - Reads like a real human wrote it, not an AI

"The AI is the same for everyone. The setup is what separates the results."

— Prashant Shinde

This guide is part of the Careers Skills AI Agility programme which takes organisations from AI aware to AI capable to AI led. If you want to bring this capability to your enterprise, reach out directly.

Partner With Us