Story · Aman's Repetitive Research Problem
He Did the Same 7 Steps Every Single Day 📋
Aman, 13, from Jaipur, had figured out how to use AI for his school projects. Every time he got a new project topic, he would: (1) ask the AI to summarise what the topic was about, (2) ask for 5 key facts, (3) ask for a simple analogy to explain it to a younger student, (4) ask for 3 questions his teacher might ask, (5) ask for a simple mind map outline, (6) ask for any Indian examples, and (7) ask for a short 150-word summary he could use as a starting point.
This took him about 20 minutes of copy-pasting and rephrasing each time. His older sister watched him and said: "Aman, you are doing the same 7 steps every time. Write them down as a template with blank spaces for the topic. Then you can just fill in the topic and run the whole thing in one go."
Aman wrote his first AI workflow. His 20-minute task took 3 minutes. And more importantly — the quality was consistent every time because the steps were always the same.
👉 This lesson explains what workflows are, how to build them, and provides 5 ready-to-use templates for common school tasks.
Section 1 of 6
🔗 What Is an AI Workflow?
An AI workflow is a repeatable, structured sequence of steps that uses AI to accomplish a larger task reliably. Instead of starting from scratch each time, you follow the same process — like a recipe.
A good workflow has three properties:
- Repeatability: You can run it again tomorrow on a different topic and get a similar quality of output.
- Reusability: You can share the template with a classmate and they can use it with their own topic.
- Quality control: Because the steps are defined, the output is consistent and easier to review and verify.
Key insight: Most beginners use AI in a conversational, unstructured way — asking one question at a time and hoping for the best. Advanced users build workflows — structured sequences that reliably produce the output they need.
Section 2 of 6
⛓️ What Is Prompt Chaining?
Prompt chaining is when the output of one prompt becomes the input to the next. Each step builds on the previous one, producing increasingly refined or complete output.
Step 1: Understand the topic
→
Step 2: Extract key points
→
Step 3: Find Indian examples
→
Step 4: Generate quiz questions
→
Final output
Why chain instead of ask everything at once?
- Asking too many things in one prompt leads to shallow, rushed answers on each part.
- Chaining lets you check and verify each step before using the output in the next step.
- You can stop a chain if an earlier step produces something inaccurate, and fix it before it propagates.
Always check each step: If Step 2 produces a factual error and you feed it directly into Step 3, the error gets built into your final output. Always glance at the output of each step before using it as input to the next.
Section 3 of 6
🔧 How to Design a Workflow
Use this 4-step process to design any AI workflow:
- Define your goal: What does the final output look like? (e.g. "A 1-page study guide on any science topic I choose")
- Break the task into steps: What has to happen, in what order, to produce that output?
- Write a prompt template for each step: Include a placeholder (like [TOPIC]) for the variable that changes each time.
- Add review checkpoints: After each step that produces a claim or fact, add a brief human check before moving on.
Template placeholder tip: Use square brackets for variables: [TOPIC], [SUBJECT], [AUDIENCE], [WORD COUNT]. This makes it easy to spot exactly what needs to be filled in when you reuse the template.
Section 4 of 6
📚 5 Ready-to-Use School Workflows
🔬 Workflow 1: Topic Deep Dive (any subject)
Step 1 "Explain [TOPIC] in simple English for a Class 8 student. Use one real Indian daily-life example."
Step 2 "Now list the 5 most important facts about [TOPIC] that a student would be tested on."
Step 3 "For each fact, write a simple analogy that a 12-year-old would understand."
Step 4 "Write 3 practice exam questions about [TOPIC] at Class 8 difficulty level, with answers."
Step 5 "Summarise the whole topic in exactly 100 words."
📝 Workflow 2: Essay Planner (Hindi, English, Social Studies)
Step 1 "I need to write a [WORD COUNT]-word essay titled '[ESSAY TITLE]'. Give me an outline with 4 sections and 2–3 bullet points per section."
Step 2 "Expand section 1 into a full paragraph. Include one real example."
Step 3 "Expand section 2 into a full paragraph." (Repeat for sections 3 and 4)
Step 4 "Write a conclusion paragraph that ties together the main argument."
Step 5 ✅ Review and edit each paragraph — do not paste AI output directly. Add your own voice and knowledge.
📅 Workflow 3: Personalised Study Plan
Step 1 "My exam is on [DATE]. I need to prepare [SUBJECTS]. I have [HOURS] free per day. Create a day-by-day study plan."
Step 2 "For [WEAKEST SUBJECT], what are the top 5 topics Class 8 students most often make mistakes on?"
Step 3 "Create a revision checklist for [WEAKEST SUBJECT] — each item should take 10–15 minutes to revise."
Step 4 "Suggest 3 active recall exercises I can use to self-test on [WEAKEST SUBJECT] without any apps."
🎨 Workflow 4: Creative Project (story, speech, poster concept)
Step 1 "I am creating a [TYPE: story / speech / poster] about [THEME] for [AUDIENCE]. Give me 3 different creative concepts with a one-sentence description of each."
Step 2 "I like concept [NUMBER]. Expand it into a full outline with a beginning, middle, and end."
Step 3 "Write the opening paragraph. It should grab the reader's attention in the first sentence."
Step 4 ✅ Continue the rest yourself using the AI outline as your guide — your own writing must be present.
🔍 Workflow 5: Research Fact Checker
Step 1 "I am researching [CLAIM]. Explain what is known about this claim based on reliable scientific consensus."
Step 2 "What would be a good source to verify this? (Give me the type of source, not a made-up URL)"
Step 3 "What is the strongest counter-argument to [CLAIM]? Explain in simple terms."
Step 4 ✅ Search one reliable Indian source (NCERT, PIB, ISRO, ICMR) yourself to verify the key claim before including it in your project.
Section 5 of 6
⚠️ Common Workflow Mistakes
| Mistake | Why it's a problem | Fix |
|---|---|---|
| Asking all steps at once in one mega-prompt | AI tries to do everything at once; answers are shallow and difficult to verify | Break into separate steps; review between each |
| Not reading the output of each step | Errors or hallucinations in Step 2 get silently built into Step 3, 4, and 5 | Always read and check each output before the next step |
| Never changing the template after it fails | If a template consistently produces vague output, the template is broken | Iterate — adjust the prompt until the quality improves |
| Submitting AI output directly | Academic dishonesty; your own thinking and understanding is not developed | Use AI to get started; always add your own voice, examples, and verification |
| Treating the workflow as the final answer | Workflows are a starting point — they produce drafts, not finished work | Every AI output needs human review, editing, and verification |
Section 6 of 6
🧠 Automation Thinking: When Should a Task Be a Workflow?
Not every AI task needs a formal workflow. Here is a simple decision rule:
- Build a workflow if: You do the same multi-step task more than once, the task has a quality bar (accuracy, format, completeness), or other people need to do the same task consistently.
- Use a one-off prompt if: You are exploring a topic for the first time, you need a quick answer, or the task is not likely to repeat.
Think like a systems designer: Every time you find yourself repeating the same sequence of prompts, that sequence is a candidate for a reusable workflow template. Save it — it is getting more valuable each time you use it.
🔗 Quiz — Lesson 4
8 questions · Click your answer · Submit for your score
1. What is the main advantage of a reusable AI workflow template?
2. In prompt chaining, what happens if Step 2 contains a factual error?
3. In a workflow template, what does [TOPIC] represent?
4. Why is it generally better to chain prompts one at a time rather than asking everything in one mega-prompt?
5. Which of these tasks is BEST suited to a reusable workflow template?
6. Aman used AI to produce a project draft in 3 minutes instead of 20. What should he ALWAYS do before submitting it?
7. In the Research Fact Checker workflow (Workflow 5), why does Step 4 say "search a reliable Indian source yourself" rather than asking the AI to do it?
8. What is "automation thinking" in the context of AI workflows?
📝 Worksheet — Build Your Own Workflow
Tip: in the print dialog, choose "Save as PDF" to download.In your notebook, complete this activity:
- Think of a task you do repeatedly for school — preparing for a test, writing a report, preparing for a debate.
- List the steps you currently do, one by one (without AI).
- For each step where AI could help, write a prompt template with a placeholder for the variable part.
- Label each step: "AI draft" or "Human review" — be honest about where you need to verify output.
- Test your workflow on one real task. Note: did the output quality improve compared to your usual approach? What would you change?
📋 Note for Parents and Teachers
What this lesson covers: The concept of AI workflows and prompt chaining, 5 ready-to-use school workflow templates, common mistakes (including academic dishonesty risks), and automation thinking as a skill. This lesson emphasises responsible AI use — always reviewing, verifying, and contributing your own understanding.
Discussion prompts:
- "What is the difference between using AI as a tool and submitting AI output as your own work? Where is the line?"
- "If you build a workflow that reliably produces good study guides in 3 minutes, are you learning less? How do you make sure you are still building knowledge?"