Zero-Shot Prompting Examples for Everyday Tasks

The Everyday Zero-Shot Pattern Most everyday tasks, such as writing an email, explaining a concept, or planning a week, do not need examples. They need a zero-shot prompt (Lesson 16) that clearly states three things:

1. What you want produced
2. The situation it is for (who, why, constraints)
3. What the result should look like Everyday prompts usually fail for one reason: the situation exists only in your head. The examples in this lesson show a weak version and an improved version of common tasks, and explain exactly which change made the difference. Each improved prompt can be reused as a pattern.

Task 1: Writing an Email Weak Prompt Write an email to my manager asking for leave. What the model must guess: dates, reason, tone, length, who covers your work, and how formal your workplace is.

Improved Prompt Write an email to my manager requesting leave.

Details:

• Dates: 10 to 12 November (3 working days)
• Reason: my sister's wedding (mention briefly)
• My tasks during that time: Arjun has agreed to handle client support
• Workplace tone: polite and professional, not overly formal

Keep it under 120 words. Include a subject line. Explanation The details list supplies the facts the email must contain; without them, the model writes placeholders such as [dates] or invents a reason. Mentioning coverage (Arjun handling support) answers the question every manager has, which makes the request more likely to be approved. Tone guidance prevents both stiff, old-fashioned phrasing and overly casual language. Subject line and length make the result ready to send. Reusable pattern: Write an email to [recipient] about [purpose]. Details: [facts]. Tone: [tone]. Length: [limit]. Include a subject line.

Task 2: Rewriting Text in a Different Tone Weak Prompt Make this sound better: "your payment didnt go through, pay again or your course will be blocked" Improved Prompt Rewrite the message below to send to a student whose payment failed.

Goals:

• Polite and reassuring, not threatening
• Clearly say what happened and what they need to do
• Keep the important fact that access will pause if payment is not completed
• Under 50 words

your payment didnt go through, pay again or your course will be blocked Possible output:

Hi! It looks like your recent payment didn't go through. Please try again using the link in your account so you can keep learning without interruption. Your course access will pause if the payment isn't completed. Need help? Just reply to this message. Explanation "Make this sound better" could mean fix grammar, make it formal, or make it friendlier. The improved prompt names the audience, the goal of the tone change, and, importantly, which fact must be kept. Without that last rule, a rewrite focused on being friendly might drop the consequence of not paying.

Reusable pattern: Rewrite the text for [audience]. Tone: [target tone]. Keep: [facts that must stay]. Length: [limit].

Task 3: Explaining a Concept Weak Prompt Explain recursion. Improved Prompt Explain recursion to someone who has just learned Python functions and loops but has never seen a function call itself.

Include:

1. A one-sentence definition
2. An everyday analogy
3. A short Python example that calculates a factorial, with each line explained
4. The most common beginner mistake with recursion

Use simple language. Keep it under 300 words. Explanation Stating what the learner already knows (functions and loops) and what they don't (self-calling functions) tells the model where to start. The numbered list of contents ensures the explanation covers definition, intuition, code, and pitfalls rather than only one of them. Asking for a specific example (factorial) keeps the code small and familiar. Reusable pattern: Explain [concept] to someone who knows [prior knowledge] but not [gap]. Include: [parts]. Length: [limit].

Task 4: Creating a Study or Work Plan Weak Prompt Make me a plan to learn Django. Improved Prompt Create a 4-week plan to learn Django.

About me:

• I know Python basics (functions, lists, dictionaries, classes)
• I can study 1 hour on weekdays and 3 hours on Saturdays
• Goal: build and deploy a simple blog with user login by the end

Format:

• One section per week with a goal for that week
• Under each week, list the topics and one small practice task
• End each week with a checkpoint: what I should be able to do Explanation Current knowledge prevents the plan from wasting time on Python basics. Available time makes the plan realistic. Without it, plans often assume several hours a day. A concrete goal (deployed blog with login) gives the plan a clear direction. Weekly checkpoints turn the plan into something you can measure progress against. Reusable pattern: Create a [duration] plan to [goal]. About me: [current level, time available]. Format: [structure].

Task 5: Brainstorming Ideas Weak Prompt Give me project ideas. Improved Prompt Suggest 8 portfolio project ideas for a beginner web developer applying for junior Django jobs in India.

For each idea, give:

• Project name
• One sentence on what it does
• The key skill it demonstrates to employers

Requirements:

• Each project should be completable in 1 to 2 weeks
• Avoid the most overused projects (to-do list, calculator, weather app)
• Include at least 2 projects that use a third-party API Explanation Purpose (portfolio for junior Django jobs) shifts suggestions toward what employers value. Excluding overused ideas pushes the model away from its most likely, and least distinctive, answers. This is a useful general technique for brainstorming: name the obvious answers you do not want. The structure per idea makes the list easy to compare. Reusable pattern: Suggest [number] ideas for [purpose] for [audience]. For each give: [fields]. Avoid: [obvious ideas]. Requirements: [constraints].

Task 6: Proofreading Without Losing Your Voice Weak Prompt Correct this. Improved Prompt Proofread the paragraph below.

• Fix only spelling, grammar, and punctuation errors.
• Do not change my wording, tone, or sentence structure unless a sentence is grammatically incorrect.
• After the corrected paragraph, list each change in the format: original → corrected

[your text] Explanation When asked to "correct" text, models often rewrite it, making it sound more polished but less like you. Limiting the scope of changes protects your voice. Asking for a list of changes lets you review every edit instead of comparing two versions line by line.

Reusable pattern: Proofread [text]. Change only [scope]. Keep [what must stay]. Then list each change as original → corrected.

Task 7: Turning Unstructured Text Into a Table Weak Prompt Organize this info. Improved Prompt Convert the notes below into a Markdown table with these columns: Course | Duration | Fee | Starts On

If a value is not mentioned, write "—". Keep the courses in the order they appear.

Python basics is 4 weeks and costs 1999, starting 1 Nov. Django course starts on 15 Nov, 6 weeks long. React is 5 weeks, fee 2999. Expected output:

Explanation Named columns define exactly what to extract. The rule for missing values prevents the model from inventing a fee for Django or a start date for React. The notes simply do not contain these facts. Preserving order keeps the table easy to check against the original notes. Reusable pattern: Convert [text] into a table with columns [list]. For missing values write [placeholder].

Task 8: Creating a Checklist Weak Prompt What should I check before launching my website? Improved Prompt Create a pre-launch checklist for a Django website going live on a Linux server with Nginx and Gunicorn.

Group the items under: Security, Performance, SEO, and Monitoring. Write each item as a short, checkable action (for example: "DEBUG is set to False"). Maximum 6 items per group. Explanation The technical environment makes items specific: Django settings, Nginx configuration, and Gunicorn are named instead of generic advice. Groups organize the checklist by area of responsibility. "Checkable action" with an example ensures each item can be marked done or not done, rather than vague advice such as "think about security". As with any technical output, verify items against official documentation before relying on them. For Django, the official deployment checklist and the python manage.py check --deploy command are useful references.

Reusable pattern: Create a checklist for [task] in [environment]. Group under [categories]. Each item must be a checkable action. Maximum [n] per group.

Task 9: Comparing Options to Make a Decision Weak Prompt Should I use MySQL or PostgreSQL? Improved Prompt I need to choose between MySQL and PostgreSQL for a new Django project.

Project details:

• An inventory system for a small retail business
• About 20 users and up to 50,000 products
• I will host it myself on a single VPS
• I have used MySQL before but never PostgreSQL

Compare the two for this specific project in a table with the columns: Factor | MySQL | PostgreSQL Cover: Django support, setup effort, learning curve for me, and features I might need. Then give a recommendation in 2 sentences. Explanation Without project details, comparisons stay generic and usually end with "it depends". This prompt provides the scale, hosting situation, and personal experience, which are the factors that actually decide the answer. Requesting specific comparison factors keeps the table relevant, and asking for a short recommendation turns the comparison into a decision.

Reusable pattern: Compare [option A] and [option B] for [specific situation]. Details: [facts]. Table columns: [list]. Factors: [list]. Then recommend in [length].

Task 10: Quick Technical Lookups Weak Prompt git undo Improved Prompt In Git, I made a commit on my local branch but have not pushed it yet. I want to undo the commit but keep my file changes so I can edit them and commit again. What command should I use? Explain what the command does and any risk. Possible output:

Use: git reset --soft HEAD~1

This moves your branch back by one commit and keeps all the changes from that commit staged, so you can edit them and commit again.

Risk: This rewrites your local history. It is safe because you have not pushed the commit. If a commit has already been pushed to a shared branch, use git revert instead, which creates a new commit that undoes the changes without rewriting history. Explanation "Undo" in Git can mean many different operations with very different consequences. The improved prompt describes the current state (committed, not pushed) and the desired result (keep the changes), which identifies one correct command. Asking about risk produces the important warning about pushed commits.

Warning: Always understand a command before running it, especially commands that delete files, rewrite history, or modify a server. Verify unfamiliar commands in the official documentation.

Reusable pattern: In [tool], my current situation is [state]. I want [result]. What command should I use? Explain what it does and any risk.

Reusable Pattern Reference Task Key Information to Include Email Recipient, purpose, facts, tone, length, subject line Rewriting tone Audience, target tone, facts to keep, length Explaining a concept Prior knowledge, knowledge gap, required parts, length Plan Current level, available time, goal, structure Brainstorming Purpose, audience, fields per idea, ideas to avoid Proofreading Scope of allowed changes, what to keep, change list Table conversion Columns, missing-value rule, order Checklist Environment, groups, checkable items, limits Comparison Your situation, factors, table format, recommendation Technical lookup Current state, desired result, ask for risks Common Mistakes in Everyday Prompting Mistake 1: Leaving Out the Personal Situation What it is: Asking general questions about personal decisions, such as "What laptop should I buy?"

Why it happens: Your needs feel obvious to you.

What it causes: Generic answers that suit an average person, not you.

How to fix it: Include your budget, use case, constraints, and preferences.

Mistake 2: Accepting Placeholder-Filled Output What it is: Sending an email or document that still contains [Your Name], [Date], or invented details.

Why it happens: The model filled gaps with placeholders or guesses because facts were missing.

What it causes: Unprofessional or incorrect messages.

How to fix it: Provide all facts in the prompt, and read the output before using it.

Mistake 3: Letting the Model Rewrite Too Much What it is: Asking to "improve" or "correct" text and receiving a full rewrite.

Why it happens: The scope of changes was not limited.

What it causes: Text that no longer sounds like you or changes your meaning.

How to fix it: State exactly what may be changed and what must stay.

Mistake 4: Sharing Sensitive Personal Information What it is: Pasting full bank details, passwords, ID numbers, or private messages into a prompt.

Why it happens: It seems convenient to include everything.

What it causes: Unnecessary exposure of sensitive data.

How to fix it: Remove or replace sensitive details; the model rarely needs them to complete the task.

Mistake 5: Trusting Technical Answers Without Checking What it is: Running commands or applying settings exactly as suggested.

Why it happens: The answer looks confident and specific.

What it causes: Data loss, broken configurations, or security problems when the answer is wrong for your situation.

How to fix it: Ask for an explanation and risks, and verify unfamiliar commands in official documentation.

Try It Yourself Exercise 1: Improve Three Everyday Prompts Improve each prompt using the patterns in this lesson:

1. Write a message to my landlord.
2. Explain APIs.
3. Help me plan my week. For each, list the information you added and which question it answers for the model.

Exercise 2: Build Your Personal Prompt Library Choose five tasks you do regularly. For each, write a reusable zero-shot prompt with placeholders in square brackets. Use each one at least twice over the next week and refine any prompt that produced output you had to edit significantly.

Exercise 3: Test the "Avoid the Obvious" Technique Ask for 10 blog post ideas about Python, first without exclusions and then with: Avoid generic topics such as "Python for beginners", "Top 10 libraries", and "Python vs Java". Compare how original the two lists are.