Prompt Engineering for Beginners India: Complete Free Course 2026

Prompt engineering is the most accessible and immediately useful AI skill you can learn in 2026. Whether you are a student preparing for JEE, a chartered accountant handling GST filings, or a developer building applications, the ability to write effective AI prompts directly determines the quality of output you get.

This is a complete, structured course. Work through each module in order, practice the exercises, and you will go from beginner to confident prompt engineer in a single sitting.

What You Will Learn

• What prompt engineering is and why it matters
• The 5 core techniques: zero-shot, few-shot, chain-of-thought, role-based, and system prompts
• How to structure prompts for different tasks
• India-specific prompt examples for real-world use cases
• Practice exercises with expected outputs
• Common mistakes and how to avoid them

Module 1: Understanding AI Models and Prompts

Before learning techniques, you need to understand what happens when you type a message to ChatGPT, Claude, or Gemini.

How AI language models work (simplified):

1. You type a prompt (your input)
2. The model processes your text and predicts the most likely helpful response
3. The model generates output token by token based on your input and its training data

The key insight: AI models do not think. They predict. Your prompt shapes what the model predicts. A vague prompt leads to generic predictions. A specific, well-structured prompt leads to precise, useful predictions.

The prompt equation:

Better Input = Better Output
Specific Context + Clear Task + Format Instructions = High-Quality Response

This is prompt engineering in one sentence: giving the AI enough context, constraints, and structure that it can only generate the response you actually want.

Module 2: Zero-Shot Prompting

Zero-shot prompting means giving the AI a task with no examples. You describe what you want, and the model uses its training to respond.

When to use zero-shot: Common tasks where the model already knows the expected format — summarization, translation, Q&A, simple analysis.

Basic zero-shot prompt:

Summarize the key provisions of India's Digital Personal Data Protection Act 2023 in 5 bullet points for a non-technical audience.

Improved zero-shot prompt (with constraints):

Summarize the key provisions of India's Digital Personal Data Protection Act 2023.
- Write exactly 5 bullet points
- Each bullet should be 1-2 sentences
- Use simple English that a Class 10 student can understand
- Focus on how this affects individual citizens, not businesses

The improved version adds format instructions, audience specification, and scope constraints. These three elements transform generic output into precisely what you need.

Practice Exercise 1: Write a zero-shot prompt that asks the AI to explain what GST (Goods and Services Tax) is to a foreign tourist visiting India. Constrain the response to 100 words, simple English, and include the current GST slab rates.

For a deeper comparison of zero-shot and few-shot approaches, read our zero-shot vs few-shot prompting guide.

Module 3: Few-Shot Prompting

Few-shot prompting is the single biggest improvement most beginners can make. Instead of describing what you want, you show the AI 2-3 examples.

Why few-shot works so well: Humans are better at recognizing patterns than writing specifications. By showing the AI examples, you communicate format, tone, and style that would take paragraphs to describe in words.

Few-shot for Indian invoice classification:

Classify the following business expenses into the correct GST category.

Example 1:
Expense: "Office rent payment - Rs 50,000/month for co-working space in Bengaluru"
Category: Renting of immovable property
GST Rate: 18%
HSN/SAC Code: SAC 9972

Example 2:
Expense: "Cloud hosting - AWS Mumbai region, Rs 15,000/month"
Category: Information technology software services
GST Rate: 18%
HSN/SAC Code: SAC 998314

Example 3:
Expense: "Company team lunch at restaurant - Rs 8,500 with AC dining"
Category: Restaurant services (with AC)
GST Rate: 5% (no ITC)
HSN/SAC Code: SAC 9963

Now classify this:
Expense: "Annual subscription to Zoho Books accounting software - Rs 12,000"

The model learns the exact output format from your examples and applies it consistently to new inputs.

Few-shot for medical history summarization:

Summarize patient history in the standard Indian medical record format.

Example:
Input: "45-year-old male, diabetic for 10 years, on Metformin 500mg BD, recent HbA1c 8.2%, complains of tingling in feet for 3 months, BMI 28"
Summary:
- Age/Gender: 45/M
- Known Case: Type 2 DM (10 years)
- Current Rx: Tab Metformin 500mg BD
- Recent Labs: HbA1c 8.2% (above target)
- Presenting Complaint: Peripheral neuropathy symptoms x 3 months
- BMI: 28 (Overweight, Indian standards)
- Flag: Suboptimal glycemic control, evaluate for diabetic neuropathy

Now summarize:
Input: "32-year-old female, first pregnancy at 28 weeks, gestational diabetes diagnosed at 24 weeks, on medical nutrition therapy, fasting glucose 105 mg/dL, post-prandial 160 mg/dL, no other comorbidities"

Practice Exercise 2: Create a few-shot prompt with 2 examples that converts informal Indian English text messages into formal business emails. Use realistic examples with Indian business context.

Module 4: Chain-of-Thought Prompting

Chain-of-thought (CoT) prompting asks the AI to show its reasoning step by step before giving a final answer. This dramatically improves accuracy on math, logic, and analysis tasks.

Why it works: When the model writes out intermediate steps, each step constrains the next, reducing errors. It is like showing your work in a math exam — you catch mistakes during the process.

Without chain-of-thought:

A shopkeeper in Mumbai buys goods worth Rs 10,000 from a manufacturer in Gujarat. GST rate is 18%. The shopkeeper sells to a customer in Mumbai at 30% markup. Calculate the GST components and final price.

With chain-of-thought:

A shopkeeper in Mumbai buys goods worth Rs 10,000 from a manufacturer in Gujarat. GST rate is 18%. The shopkeeper sells to a customer in Mumbai at 30% markup. Calculate the GST components and final price.

Think step by step:
1. First, identify whether this is IGST or CGST+SGST based on the transaction type
2. Calculate the purchase price including GST
3. Calculate the selling price with markup
4. Calculate the applicable GST on the sale
5. Determine the input tax credit
6. Calculate the net GST payable
7. State the final price to the customer

Show all calculations clearly.

The step-by-step structure forces the model to correctly identify inter-state (IGST) vs intra-state (CGST+SGST) transactions, which it might confuse without structured reasoning.

Chain-of-thought for legal analysis:

Analyze whether the following situation constitutes wrongful termination under Indian labour law. Think through each element:

Situation: An IT company in Hyderabad terminated a pregnant employee during her probation period, citing "performance issues" with no prior written warnings.

Step 1: Identify applicable laws (Maternity Benefit Act, Industrial Disputes Act, state-specific Shops and Establishments Act)
Step 2: Analyze the protection status of the employee
Step 3: Evaluate the employer's stated reason against legal requirements
Step 4: Consider relevant court precedents
Step 5: Provide a clear conclusion with recommended actions

For a comprehensive deep dive into chain-of-thought and advanced reasoning techniques, see our chain-of-thought prompting guide.

Module 5: Role-Based Prompting

Role-based prompting assigns the AI a specific identity, expertise, and perspective. This focuses the model's responses within a particular domain.

Basic role prompt:

You are a senior SEBI-registered investment advisor with 20 years of experience in the Indian equity market. A 28-year-old software engineer earning Rs 15 LPA in Bangalore asks you about starting to invest. They have Rs 5 lakh in savings and no existing investments.

Provide specific, actionable advice following SEBI guidelines. Include:
- Asset allocation recommendation
- Specific instrument types (not stock names)
- Tax-saving options under Section 80C
- Monthly SIP recommendations
- What to avoid as a first-time investor

Role + constraints for education:

You are an experienced CBSE Physics teacher who has taught Class 12 for 15 years. A student is struggling with electromagnetic induction.

Explain Faraday's Law of Electromagnetic Induction:
- Use simple Hindi-English mixed language (Hinglish) where it helps clarity
- Include a real-world example they can see in their home
- Connect it to the NCERT textbook Chapter 6 structure
- End with 2 NCERT-style numerical problems with solutions
- Difficulty: Board exam level

This technique is especially powerful when combined with system prompts, which let you set the role once for an entire conversation.

Module 6: System Prompts and Persistent Instructions

System prompts define the AI's behavior for an entire conversation. Instead of repeating your role instructions in every message, you set them once.

In ChatGPT: Go to Settings → Personalization → Custom Instructions In Claude: Use the system prompt field in the API, or Projects in Claude.ai In Gemini: Use the system instruction field

Example system prompt for a CA practice:

You are an AI assistant for a Chartered Accountancy firm in India. Follow these rules:

ROLE: Senior tax consultant specializing in Indian direct and indirect taxation
KNOWLEDGE: Income Tax Act 1961, GST Act 2017, Companies Act 2013, SEBI regulations
LANGUAGE: Professional English with standard Indian accounting terminology
FORMAT: Always structure responses with clear headings, relevant section numbers, and practical next steps

RULES:
- Always cite specific sections, rules, or notifications when referencing tax law
- Clearly state when information might be outdated and recommend verifying with current circulars
- For calculations, show complete working with each step
- If a query requires professional judgment, state "This requires professional review" and explain why
- Never provide advice on tax evasion — only legal tax planning and optimization

For a complete guide on writing effective system prompts with templates, read our system prompts guide.

Module 7: Combining Techniques

Real-world prompt engineering uses multiple techniques together. Here is a complex prompt that combines role, chain-of-thought, and format constraints.

Combined prompt for legal document analysis:

You are a corporate lawyer specializing in Indian contract law with 15 years of experience at a top-tier law firm.

I will share a vendor agreement. Analyze it step by step:

Step 1: Identify all parties and their obligations
Step 2: Flag clauses that are unusual, one-sided, or potentially unenforceable under Indian Contract Act 1872
Step 3: Check for missing standard clauses (indemnity, limitation of liability, governing law, dispute resolution)
Step 4: Assess IP ownership and assignment clauses
Step 5: Review termination clauses and notice periods
Step 6: Provide a risk rating (Low/Medium/High) for each flagged issue

Format your response as a table:
| Clause | Section | Issue | Risk | Recommendation |

End with a 3-line executive summary suitable for a CTO who is not a lawyer.

Module 8: Common Mistakes and How to Fix Them

| Mistake | Example | Fix | |---------|---------|-----| | Too vague | "Tell me about Indian tax" | "Explain the 5 GST slab rates in India with 2 product examples for each slab" | | No format specified | "Analyze this data" | "Analyze this data and present findings as a numbered list with one insight per point" | | Asking for too much at once | "Write a complete business plan" | Break into sections: first market analysis, then financial projections, then marketing strategy | | Not specifying audience | "Explain machine learning" | "Explain machine learning to a Class 10 student who knows basic math but no programming" | | Ignoring context window | Pasting 100 pages of text | Summarize key sections first, then ask specific questions about each section |

Practice Exercises: Put It All Together

Exercise 1 — Zero-shot: Write a prompt that asks the AI to create a study schedule for a JEE Main aspirant with 6 months remaining, covering Physics, Chemistry, and Math.

Exercise 2 — Few-shot: Create a prompt with 3 examples that converts Indian addresses from informal format to India Post standard pin code format.

Exercise 3 — Chain-of-thought: Write a prompt that calculates whether it is better to take the old tax regime or new tax regime for a person earning Rs 12 LPA with Rs 1.5 lakh in 80C investments and Rs 50,000 in NPS.

Exercise 4 — Combined: Write a system prompt for an AI teaching assistant that helps NEET aspirants study Biology, uses NCERT as the primary reference, explains in Hinglish, and includes mnemonics for difficult topics.

Where to Go From Here

This course covers the fundamentals. To continue building your prompt engineering skills:

1. Practice daily — Use AI tools for real tasks and iterate on your prompts
2. Learn advanced techniques — Advanced prompting with ReAct and Tree-of-Thought covers research-backed methods
3. Explore tool-specific prompts — Our prompts by AI tool guide shows what works best in ChatGPT vs Claude vs Gemini
4. Build a prompt library — Save your best prompts and refine them over time using our prompt templates library
5. Consider a career in prompt engineering — How to become a prompt engineer in India covers salary, skills, and job opportunities

Prompt engineering is not a one-time learning exercise. As AI models improve and new capabilities emerge, the techniques evolve. But the fundamentals you learned here — specificity, structure, examples, and reasoning — will remain the foundation of effective AI interaction for years to come.