Introduction
In Part 1 of our AI Prompting Series, we introduced the CLEAR Framework — a simple 5-step method to write effective prompts (Context, Length & Format, Expectation, Audience, Restrictions).
If you haven't read Part 1 yet, we recommend reading it first. The CLEAR Framework is the foundation.
Now, let's level up.
Why Do Different Prompting Methods Exist?
Because writing a good prompt is only half the battle. The other half is choosing the right prompting method for your task.
Think of it like this: A doctor doesn't use the same tool for every patient. A stethoscope is for the heart, an X-ray is for bones, and an MRI is for detailed internal scans. Similarly, different prompting methods are designed for different types of tasks.
Using the wrong method is like using a stethoscope to check for a fracture — you'll get an answer, but it won't be the right one.
Here's a simple rule to remember:
Simple task → Simple method.
Complex task → Advanced method.
High-stakes decision → Multi-step method.
In this blog, we cover 10 proven prompting methods — each explained in simple language with real-world examples, best-use industries, and a comparison table to help you choose the right one.
The 10 Prompting Methods — Quick Overview
# | Method | Difficulty | Best For | One-Line Summary |
1 | Zero-Shot | Beginner | Quick questions | Ask directly — no examples needed |
2 | Few-Shot | Beginner | Formatting, classification | Show examples first, then ask |
3 | Role Prompting | Beginner | Expert-level answers | Tell AI “who” to be |
4 | Chain-of-Thought | Intermediate | Math, logic, reasoning | Ask AI to think step by step |
5 | Tree-of-Thoughts | Intermediate | Strategic decisions | Explore multiple paths, choose best |
6 | Self-Consistency | Intermediate | Accuracy verification | Multiple answers, pick most common |
7 | ReAct | Advanced | Research, fact-checking | Think, act, observe, repeat |
8 | Meta Prompting | Advanced | Prompt improvement | AI writes better prompts for you |
9 | Prompt Chaining | Advanced | Complex workflows | Break big tasks into linked steps |
10 | Generate Knowledge | Intermediate | Research, analysis | AI studies first, then answers |
Method 1: Zero-Shot Prompting
Difficulty: Beginner | Best for: Quick factual answers, simple translations, basic summaries
You ask the AI a question directly — no examples, no templates. Just ask and the AI figures it out on its own.
Simple Analogy: Walking into a restaurant and saying, “Give me something good to eat.” The chef uses their own judgment.
Example
Prompt: What is the due date for filing GSTR-3B for the month of January 2026?
Best Industries: All industries — this is the default starting point for everyone.
Method 2: Few-Shot Prompting
Difficulty: Beginner | Best for: Consistent formatting, classification, tone matching
You give the AI 2–3 examples of what you want, and then ask your actual question. The AI learns from your examples and follows the same pattern.
Simple Analogy: Showing your junior 2–3 sample letters before asking them to write a new one. They follow the same style.
Example
Prompt: Here are examples of how I classify client queries: “I received a notice under Section 148” → Income Tax Notice “My GST registration was cancelled” → GST Compliance Now classify: “I want to claim ITC on capital goods”
Best Industries: Legal, Finance, Marketing, Education.
Method 3: Role (Persona) Prompting
Difficulty: Beginner | Best for: Expert-level responses, domain-specific advice
You tell the AI to “become” a specific expert before answering. This shapes the tone, depth, and perspective of the response.
Simple Analogy: Instead of asking a general doctor about a heart problem, you specifically ask a cardiologist.
Example
Prompt: You are a senior CA with 20 years of experience. A salaried client earning ₹18 lakh asks whether to choose old regime or new regime for FY 2025-26. Explain with comparison.
Best Industries: Healthcare, Legal, Finance, Education, Marketing.
Method 4: Chain-of-Thought (CoT) Prompting
Difficulty: Intermediate | Best for: Math, tax computations, logical reasoning, audit analysis
You ask the AI to show its reasoning step by step before giving the final answer — just like a CA shows workings in a tax computation.
Simple Analogy: In a maths exam, the teacher says “show your working.” Chain-of-Thought is exactly that — for AI.
Example
Prompt: A business has turnover of ₹85 lakh in FY 2025-26. Do they need GST registration? Think step by step.
Best Industries: Finance & Accounting, Engineering, Education, Legal.
Method 5: Tree-of-Thoughts (ToT) Prompting
Difficulty: Intermediate | Best for: Strategic decisions, comparing alternatives
The AI explores multiple possible paths, evaluates each one, and then chooses the best option. Like a decision tree.
Simple Analogy: Comparing 3 different vacation options before deciding. ToT does the same for AI reasoning.
Example
Prompt: A business owner has ₹20 lakh to invest. Explore three strategies — FD, Mutual Funds, Commercial Property. Analyse pros, cons, and expected returns. Recommend for moderate-risk investor.
Best Industries: Management Consulting, Corporate Strategy, Investment, Real Estate.
Method 6: Self-Consistency Prompting
Difficulty: Intermediate | Best for: High-accuracy tasks, verification, compliance
Ask the AI to generate multiple answers, then pick the one that appears most consistently. If 3 out of 4 agree, that’s probably correct.
Simple Analogy: Asking 5 different CAs about a transaction’s tax treatment. If 4 agree, you trust that answer.
Example
Prompt: Generate three independent tax calculations for income of ₹15L salary + ₹2L house property + ₹80K FD interest under new regime FY 2025-26. Then identify the most consistent answer.
Best Industries: Healthcare, Legal, Auditing, Financial Advisory.
Method 7: ReAct (Reason + Act) Prompting
Difficulty: Advanced | Best for: Research, fact-checking, real-time data
The AI alternates between thinking (reasoning) and doing (searching/acting). It’s a loop: Think → Act → Observe → Think again.
Simple Analogy: A CA student researching a case — they think about the section, look up the circular, read it, then form an opinion.
Example
Prompt: Find the latest CBDT circular on extended ITR due date for AY 2025-26 audit cases. Think about what info you need, search, then summarise.
Best Industries: Journalism, Legal Research, Academic Research, Market Research, Compliance.
Method 8: Meta Prompting
Difficulty: Advanced | Best for: Improving prompts, creating templates, automation
Instead of writing the prompt yourself, you ask the AI to write or improve the prompt for you. The AI designs a better prompt.
Simple Analogy: Instead of writing your own exam paper, you ask an experienced teacher to design it. They know how to frame better questions.
Example
Prompt: I want to write a professional email about a missed GST deadline. Create the perfect prompt for this task with all context, format, and restrictions.
Best Industries: IT & Software, Corporate Training, Content Marketing, Customer Support, AI Development.
Method 9: Prompt Chaining
Difficulty: Advanced | Best for: Complex multi-step projects, report generation
Break a large task into smaller steps. Output of one prompt becomes input for the next — like an assembly line.
Simple Analogy: Filing ITR is not one task — it’s a chain: Gather docs → Compute income → Calculate tax → Prepare ITR → Verify and file.
Example
Step 1: List all income heads for an individual with salary, house property, and FD income.
Step 2: Compute gross total income for ₹12L salary + ₹1.8L rent + ₹50K FD interest.
Step 3: Calculate tax under old and new regime and recommend better option.
Best Industries: Accounting & Audit, Legal, Content Creation, Software Development, Project Management.
Method 10: Generate Knowledge Prompting
Difficulty: Intermediate | Best for: Unfamiliar topics, deep research, comprehensive reports
Before asking the AI to answer, you first ask it to generate background knowledge on the topic. It’s like studying before the exam.
Simple Analogy: Before advising on international taxation, a CA reads up on DTAA and FEMA first. This method makes the AI “study” first.
Example
Step 1: Generate a comprehensive summary of key changes in the Income Tax Act, 2025 effective from April 2026.
Step 2: Based on that knowledge, write a 500-word client advisory for a small business owner in Ahmedabad.
Best Industries: Research & Academia, Consulting, Policy Analysis, Healthcare, Financial Planning.
Industry-Wise Best Methods
Industry | Top 3 Methods | Why These Work Best |
Chartered Accountants | CoT, Few-Shot, Prompt Chaining | Tax needs step-by-step logic; compliance needs formatting; audits need workflows |
Lawyers & Advocates | Role, ReAct, Generate Knowledge | Legal opinions need expertise; research needs search; new laws need study |
Doctors & Healthcare | Self-Consistency, CoT, Role | Diagnosis accuracy critical; medical reasoning needs logic; specialist perspective |
Business Owners | ToT, Zero-Shot, Prompt Chaining | Decisions need comparison; daily queries simple; projects need steps |
Marketing Professionals | Few-Shot, Role, Meta Prompting | Brand voice needs examples; creative needs persona; campaigns need optimisation |
Teachers & Educators | CoT, Few-Shot, Generate Knowledge | Teaching needs step-by-step; exams need format; new syllabus needs research |
Software & IT | Prompt Chaining, Meta, ReAct | Development is multi-step; automation needs prompts; debugging needs research |
Common Mistakes to Avoid
Mistake | What Happens | Solution |
Zero-Shot for complex reasoning | Vague or incorrect answers | Switch to Chain-of-Thought |
Too many examples in Few-Shot | AI follows wrong pattern | Limit to 2–3 clear examples |
Wrong role assignment | Off-topic responses | Match role to your domain |
Advanced method for simple Q | Wasted time, long response | Start with Zero-Shot, upgrade if needed |
Not verifying critical answers | Errors in compliance/calculations | Use Self-Consistency for high-stakes |
What’s Next? (Coming in Part 3)
In the next blog, we will dive deep into each method with detailed, step-by-step tutorials — including multiple real-world examples from accounting, legal, medical, and business domains. We’ll also share copy-paste prompt templates you can use immediately.
Stay tuned!
About the Author
Himanshu Majithiya & Co. is a Chartered Accountants firm established in 2007, based in Ahmedabad, Gujarat. The firm provides services in Income Tax, GST, Company Audit, FFMC Compliance (RBI), and AI & Workflow Automation.
Website: www.himanshumajithiya.com | Contact: +91 98795 03465 | info@himanshumajithiya.com
Disclaimer: This blog is published for educational and informational purposes only. It does not constitute professional advice. The content is based on publicly available research papers, AI documentation, and industry best practices. Readers are advised to consult qualified professionals for specific advice. Published in compliance with ICAI advertising guidelines.
© 2026 Himanshu Majithiya & Co. All rights reserved.