Series Recap: The Complete Journey
Part | Blog Title | Status |
Part 1 | The Art of Prompting — CLEAR Framework | Completed |
Part 2 | 10 Prompting Methods That Get Results | Completed |
Part 3A | Mastering the Basics — 5 Beginner/Intermediate Methods | Completed |
Part 3B | Advanced Prompting — 5 Power Methods (This Blog) | You are here |
These advanced methods build on the basics. If you haven’t read Part 3A yet, we recommend reading it first.
Method 6: Tree-of-Thoughts (ToT) — Explore All Options Before Deciding
What it is: AI explores multiple possible solutions side by side, evaluates each independently, and selects the best option with reasoning.
In one line: “Don’t go with the first idea — compare all options, then pick the best.”
When to Use It
Use ToT When... | Don’t Use When... |
Making a strategic decision with multiple options | Question has a single factual answer |
Need to compare alternatives with pros/cons | Need a quick, simple response |
Best answer requires weighing trade-offs | Task doesn’t involve choices |
Example 1: For CAs — Firm Growth Strategy
Scenario: Grow revenue by 40% — compare three growth strategies.
Prompt: I run a CA firm in Ahmedabad (est. 2007, revenue ~Rs. 30L). I want to grow to Rs. 42L in FY 2026-27. Explore 3 strategies. For each: actions, investment, timeline, expected revenue, top 3 risks, top 3 advantages. A: Add GSTR-9/9C as dedicated service line B: Launch AI & Automation consulting for other firms Compare in table. Recommend best for risk-moderate owner with Rs. 3L upfront budget.
Example 2: For Business Owners — Office Space Decision
Scenario: Business growing, need to decide on next office setup.
Prompt: Textile export business, 8 employees, current office too small. Explore 3 options. Analyse: 3-year total cost, productivity impact, tax benefits, flexibility, hidden costs. A: Larger rented office, SG Highway, Rs. 75K/month B: Purchase commercial office, Thaltej, Rs. 55L C: Premium co-working space, Rs. 50K/month Compare with 3-year cost table. Recommend for 20-30% annual growth.
Example 3: For Marketing — Campaign Strategy
Scenario: Rs. 2L quarterly budget, need optimal channel mix.
Prompt: CA firm marketing, Rs. 2L/quarter, goal: 50 qualified leads/quarter. Explore 3 strategies. Analyse: budget split, expected leads, CPL, time to results, team effort, sustainability. A: Google Ads focused (80% on search) B: Content marketing focused (LinkedIn + Instagram + YouTube) C: Referral + partnership focused (client bonuses, events) Compare and recommend for sustainable long-term lead generation.
Copy-Paste Template
Template: I need to decide about [situation]. Explore [3] options. For each analyse: - [Criteria 1] - [Criteria 2] - [Criteria 3] Option A: [Description] Option B: [Description] Option C: [Description] Compare in table. Recommend for [your profile/constraints].
Pro Tips
Tip | Why It Helps |
Always define 3+ options, not just 2 | Three forces genuine comparison, two creates bias |
Specify evaluation criteria upfront | Prevents AI from using random criteria |
Include your constraints in final ask | Sharpens the recommendation |
Ask for comparison table at the end | Tables reveal trade-offs at a glance |
Method 7: Self-Consistency — Trust But Verify
What it is: AI solves the same problem multiple times using different reasoning approaches, then compares to find the most consistent (likely correct) answer.
In one line: “Solve it three ways. If they all agree, trust the answer.”
When to Use It
Use Self-Consistency When... | Don’t Use When... |
Calculations where errors are costly | Question is subjective or creative |
Compliance or legal decisions | Speed matters more than accuracy |
Want to cross-check AI’s reasoning | Answer is a straightforward fact |
Wrong answer has financial/legal consequences | Brainstorming or exploring ideas |
Example 1: For CAs — TDS Verification
Scenario: Verify TDS rate and amount for a complex contractor payment.
Prompt: Company pays Rs. 8.5L to resident contractor (not individual/HUF) for works contract (labour + material). PAN provided. Verify TDS using 3 approaches: 1. Section-based: Identify applicable section, rate, compute 2. Nature-of-payment: Analyse works contract nature, determine section 3. Threshold-based: Check thresholds, verify applicability Show complete working separately. Compare. If they match — confirm. If not — explain why.
Example 2: For Business Owners — Loan EMI Verification
Scenario: Bank offered loan, verify the EMI calculation.
Prompt: Loan: Rs. 25L, 11.5% reducing balance, 5 years, processing fee 1.5%. Bank says EMI = Rs. 55,100. Verify using 3 methods: 1. EMI Formula: P × r × (1+r)^n / [(1+r)^n - 1] 2. Total Interest Method: Calculate total interest, adjust for reducing balance 3. Year-by-Year: Check Year 1, 2, 5 to verify balance reaches zero Is bank’s Rs. 55,100 correct? Also calculate effective cost including processing fee.
Example 3: For Marketing — Conversion Funnel Verification
Scenario: Team claims 4.2% conversion rate. Verify.
Prompt: Campaign data: 45K impressions, 2700 clicks, 2200 page views, 180 forms, 95 qualified leads, 12 clients signed. Verify “4.2% conversion rate” using 3 approaches: 1. Click-to-Client rate 2. Lead-to-Client rate 3. Full funnel stage-by-stage Which rate is the team likely reporting? Which is most meaningful?
Copy-Paste Template
Template: [Problem with all data] Verify using 3 independent approaches: Approach 1 — [Name]: [Method] Approach 2 — [Name]: [Method] Approach 3 — [Name]: [Method] Show complete working separately. Compare. If match — confirm. If not — explain which is most reliable.
Pro Tips
Tip | Why It Helps |
Name each approach clearly | Helps AI keep them genuinely separate |
Ask for “independent” approaches | Prevents AI copying same logic three times |
Always request comparison at end | Comparison step is where errors get caught |
Combine with CoT for each approach | Step-by-step + triple-check = maximum verification |
Method 8: ReAct (Reason + Act) — The Research Assistant
What it is: AI alternates between thinking (reasoning) and doing (searching/acting) in a loop: Think → Act → Observe → Repeat.
In one line: “Think about what you need, go find it, then think again with what you found.”
Example 1: For CAs — Scheme Research
Scenario: Research Vivad Se Vishwas Scheme for a client with Rs. 12L pending demand.
Prompt: Client has Rs. 12L demand under Section 143(3), AY 2019-20, appeal pending at CIT(A). Research process: Thought 1: Key provisions of Vivad Se Vishwas 2024? Action 1: Research eligibility, payment, deadline, benefits Thought 2: Is client eligible (appeal at CIT(A) level)? Action 2: Check CIT(A) level eligibility criteria Thought 3: Settlement amount vs fighting the case? Action 3: Calculate scheme payment vs full demand Thought 4: Latest deadline or CBDT extensions? Action 4: Search for latest notification Final Answer: Recommendation with exact payment, deadline, and application steps.
Example 2: For Business Owners — Competitor Analysis
Scenario: Launching premium A2 ghee brand, need competitive landscape.
Prompt: Launching A2 Gir Cow Ghee from Ahmedabad. Thought 1: Top premium ghee brands in India? Action 1: Research 5-7 brands with prices, sizes, claims Thought 2: Price range for 500ml and 1L? Action 2: Check prices on Amazon, BigBasket Thought 3: Certifications competitors use? Action 3: Research claims (A2, organic, bilona, FSSAI, GI tag) Thought 4: Customer complaints in reviews? Action 4: Check Amazon reviews for gaps Final Answer: Pricing strategy, positioning angle, 3 differentiators. Competitor comparison table.
Example 3: For Marketing — Trending Content Research
Scenario: Research what’s going viral in Indian finance/tax niche on Instagram.
Prompt: Research viral content in Indian finance/tax education niche. Thought 1: Top creators in this niche? Action 1: Identify 5-7 popular creators, followers, style Thought 2: Which Reel types get most views? Action 2: Analyse patterns — educational, myth-busting, news? Thought 3: What hooks and formats work? Action 3: Identify opening hooks, video lengths, styles Thought 4: Trending topics March 2026? Action 4: Check — FY closing, advance tax, new tax code? Final Answer: 2-week calendar with 8 Reel ideas (hook, topic, format, time to create).
Copy-Paste Template
Template: I need research on [topic]. Thought 1: [What to find first?] Action 1: [Search for specific info] Observation 1: [Summarise] Thought 2: [What’s next based on Obs 1?] Action 2: [Search] Observation 2: [Summarise] Thought 3: [What else?] Action 3: [Search] Observation 3: [Summarise] Final Answer: [Actual question]. Present as [format].
Pro Tips
Tip | Why It Helps |
Structure 3-4 Thought-Action cycles | Fewer = shallow research |
Make each Action specific and different | “Research pricing” and “Research reviews” — not “research more” |
Use with AI tools that have web access | ReAct shines with actual internet search |
End with clear “Final Answer” instruction | Without this, AI may list observations without synthesis |
Method 9: Meta Prompting — Let AI Write Your Prompts
What it is: Ask the AI to create, improve, or optimise prompts for you. Describe what you need and AI designs a better prompt.
In one line: “I don’t know how to ask this perfectly — so you write the question for me.”
Example 1: For CAs — Sensitive Client Email
Scenario: Need email about GST penalty but unsure how to frame the prompt.
Prompt: I need an email to a long-term client about GSTR-3B non-filing for Oct-Dec 2025 (late fee Rs. 15K + interest Rs. 8K). Their delay caused it. Email should: state facts without blame, mention consequences, request Jan-Mar data immediately, subtly convey responsibility, maintain relationship. Create the perfect prompt I can use with any AI tool. Include: role, context, tone, format, restrictions. Also explain WHY you structured it that way — I want to learn.
Example 2: For Business Owners — Investor Pitch
Scenario: Need investor pitch deck but don’t know how to prompt for it.
Prompt: I’m a textile exporter in Ahmedabad. Need pitch for Rs. 1 crore investment (15% equity). Turnover Rs. 2.5 crore, 12% net margin. Write the perfect prompt to generate: 10-slide deck outline, key financial metrics, compelling story, top 5 investor Q&As. Make it detailed enough that a first-time AI user gets great results.
Example 3: For Marketing — Content Calendar Prompt
Scenario: Previous prompts gave generic results. Need better prompt.
Prompt: My AI content calendar prompts always give generic “post about tax tips” results. Create an optimised prompt for March 2026 content calendar: - LinkedIn (3x/week) + Instagram (5x/week) - Target: SMB owners Ahmedabad - Key dates: FY closing, advance tax deadline, GST annual return - Differentiator: AI & Automation services - Budget: Rs. 10K/month Force AI to give specific titles, hooks, format, hashtags — not vague categories. Also explain what made my previous prompts fail.
Copy-Paste Template
Template: I want AI to help me [goal]. My situation: [context 1], [context 2], [context 3] Output needed: [req 1], [req 2], [req 3] Create the perfect prompt for any AI tool. Include: expert role, context, format, tone, restrictions. Also explain why you structured it that way.
Pro Tips
Tip | Why It Helps |
Describe the goal, not the prompt | “I want an investor pitch” beats “write a prompt about pitches” |
Tell AI what went wrong before | “Previous prompts gave generic results” helps AI fix the issue |
Ask for the “why” explanation | Learning why makes you better at prompting forever |
Save good meta-prompts as templates | Build a reusable library for repeated tasks |
Method 10: Prompt Chaining — The Assembly Line
What it is: Break a large task into smaller linked prompts where output of one becomes input for the next. Each link does one job well.
In one line: “Don’t try everything in one prompt — build an assembly line.”
When to Use It
Use Chaining When... | Don’t Use When... |
Task has multiple distinct stages | Task can be done in one prompt |
Each stage needs different focus | Adding steps = unnecessary complexity |
Output of one step feeds the next | All information is available upfront |
Quality matters more than speed | Simple one-time question |
Example 1: For CAs — Audit Report Workflow
Scenario: Prepare a management letter for a manufacturing client.
Step 1: List top 15 audit observation areas for a manufacturing company (Rs. 15 crore turnover). Categorise: Inventory, Revenue, Expenses, Statutory Compliance, Internal Controls.
Step 2: Based on above, draft observations for 5 issues we found: 1. Inventory difference Rs. 3.2L 2. 12 invoices (Rs. 8.5L) recorded in wrong period 3. TDS Rs. 2.1L deducted but not deposited on time 4. No approval process for purchases >Rs. 50K 5. 8 unreconciled bank items (Rs. 4.7L) For each: finding, risk, recommendation. Formal audit language.
Step 3: Draft complete management letter to Board of Directors. Include: formal opening, numbered observations with severity ratings (H/M/L), summary table, closing. From “Himanshu Majithiya & Co., Chartered Accountants.”
Example 2: For Business Owners — Business Plan
Scenario: Complete business plan for a cloud kitchen in Ahmedabad.
Step 1: Market research: Cloud kitchen for healthy meals, Ahmedabad. Target: gym-goers age 22-40. Cover: market size, 5 competitors, pricing, delivery platforms, trends.
Step 2: Based on above: 12-month financial projections. Setup costs, monthly fixed/variable costs, revenue (3 scenarios), break-even, month-by-month P&L.
Step 3: Based on Steps 1+2: Operations plan. Daily workflow, menu structure, suppliers, quality checklist, FSSAI licensing, tech stack, staffing.
Step 4: Based on all above: 1-page executive summary for investors/bank loan. Business concept, market opportunity, competitive advantage, financials, funding need.
Example 3: For Marketing — Full Campaign Launch
Scenario: Launch AI & Automation service marketing campaign.
Step 1: Identify 3 ideal customer personas for “AI & Automation for Businesses” from our CA firm. Demographics, pain points, discovery channels, buying triggers.
Step 2: Based on personas: Messaging framework. Core value proposition, 3 benefit statements, objection handling (top 5), proof points, 3 tagline options.
Step 3: Based on Steps 1+2: 30-day content plan. LinkedIn (3x/week) + Instagram (4x/week). Each post: date, platform, format, topic, hook, message, CTA, hashtags.
Step 4: Based on Steps 1+2: Google Ads copy (3 campaigns × 3 variations each) + LinkedIn sponsored content (3 variations). Headlines, descriptions, CTAs.
Copy-Paste Template
Template: --- STEP 1: [Stage Name] --- [First stage prompt. Output feeds Step 2.] --- STEP 2: [Stage Name] --- Based on Step 1 output, [second stage task]. --- STEP 3: [Stage Name] --- Based on Steps 1+2, [third stage task]. --- STEP 4: [Final Stage] --- Based on everything above, [final deliverable]. Make it [polished/complete/ready].
Pro Tips
Tip | Why It Helps |
Keep chains to 3-5 steps | More than 5 = hard to manage, AI loses context |
Review output at each step | Catch errors before they cascade |
Use “Based on the above” to link steps | Tells AI to use previous output as context |
Run all steps in same chat session | AI remembers previous outputs |
If a step fails, only re-run that step | Biggest advantage over one giant prompt |
Master Comparison: All 10 Methods
# | Method | Difficulty | Best For | Key Trigger |
1 | Zero-Shot | Beginner | Quick questions | “Simple answer needed” |
2 | Few-Shot | Beginner | Consistent formatting | “Make it like these examples” |
3 | Role Prompting | Beginner | Expert perspective | “Need specialist advice” |
4 | Chain-of-Thought | Intermediate | Calculations, logic | “Show the working” |
5 | Generate Knowledge | Intermediate | New/complex topics | “Research first” |
6 | Tree-of-Thoughts | Intermediate | Strategic decisions | “Compare my options” |
7 | Self-Consistency | Intermediate | High-accuracy | “Must be 100% sure” |
8 | ReAct | Advanced | Live research | “Find latest info” |
9 | Meta Prompting | Advanced | Prompt improvement | “Help me ask better” |
10 | Prompt Chaining | Advanced | Multi-step projects | “Big project, many stages” |
The Prompting Power Matrix: Combining Methods
Combination | What It Does | Best For |
Role + CoT | Expert-level reasoned calculations | Tax computations, legal analysis |
Role + ToT | Expert comparing strategies | Investment decisions, planning |
Few-Shot + Chaining | Consistent format, multi-step | Report series, content calendars |
Meta + Role | AI creates expert-persona prompts | Building prompt libraries |
ReAct + Self-Consistency | Researched and verified answers | Compliance, due diligence |
Knowledge + ToT | Deep research then compare | Market entry, policy analysis |
CoT + Self-Consistency | Triple-verified calculations | Audit, financial modelling |
Your AI Prompting Journey: Complete Roadmap
Stage | What to Learn | Blog |
Foundation | How to write a clear prompt | Part 1 — CLEAR Framework |
Awareness | Which methods exist | Part 2 — 10 Methods Overview |
Proficiency | Master 5 basic methods | Part 3A — Mastering the Basics |
Mastery | Master 5 advanced methods | Part 3B — This Blog (Series Finale) |
Series Complete: You Are Now an AI Prompting Pro
Congratulations — you’ve completed the entire AI Prompting Series! Here’s what you’ve accomplished:
What You Learned | Where |
How to write clear, effective prompts using the CLEAR Framework | Part 1 |
Overview of 10 prompting methods and when to use each | Part 2 |
Deep mastery of 5 beginner/intermediate methods with templates | Part 3A |
Deep mastery of 5 advanced methods with templates | Part 3B |
You now have 40+ real-world examples and 10 copy-paste templates covering every prompting method — from simple Zero-Shot questions to complex Prompt Chaining workflows. Whether you are a Chartered Accountant computing taxes, a business owner making strategic decisions, or a marketing professional planning campaigns, you have the tools to get exceptional results from any AI tool.
Remember: The best way to master prompting is to practice daily. Open your favourite AI tool, pick a method from this series, and start using it on your real work. Every prompt you write makes you better.
If you or your team need hands-on training or customised AI workflow automation for your business, we’d love to help.
About the Author
Himanshu Majithiya & Co. is a Chartered Accountants firm established in 2007, based in Ahmedabad, Gujarat. Services: 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, financial, legal, or tax advice. Examples are illustrative and may not reflect exact provisions applicable to your situation. Consult qualified professionals before making decisions. Published in compliance with ICAI advertising guidelines.
© 2026 Himanshu Majithiya & Co. All rights reserved.