Real-world examples

Reading time: ~15 min · Prereqs: Lessons 01–06 · 3 interactive demos inside

Why concrete examples beat abstract advice

You've learned what AI is, how to access it, and how to prompt it well. Now the question everyone eventually asks is: "But what do I actually use it for?"

This lesson is the answer. Not a list of categories, but real, specific, copy-and-use examples across five domains: everyday life, work, creative, learning, and coding. By the end, you'll have a mental library of prompts you can reach for whenever you need them.

The common thread across all five categories is the same: AI is a force multiplier for tasks you could do yourself, but would rather do faster — especially first drafts and "getting started." Understanding this frame prevents both over-reliance (treating AI as an oracle) and under-use (never trying it because you're not sure if it "counts" as a valid use case).

Category 1 — Everyday life

The most underestimated use case. The tasks we do every day — writing emails, reading documents, making plans, understanding confusing things — consume real time and mental energy. None of them require AI to be "smart" in any deep sense. They require drafting speed, patience, and the ability to explain things clearly. AI is excellent at all three.

Turning messy thoughts into clean emails

Most people write emails in three stages: first, they get the thoughts out in a messy, informal way; then they rewrite it into something readable; then they second-guess the tone and rewrite again. AI collapses this into one step. You dump your thoughts (in any order, in any style), describe who you're writing to and what you want them to feel, and ask for a draft. You read it, adjust anything that doesn't sound like you, and send.

This works for difficult emails especially well. Chasing an overdue invoice, responding to a complaint, telling a colleague something awkward — the hard part of these emails isn't the writing, it's navigating the tone. You can tell AI "be firm but not aggressive" or "be warm but not apologetic" and it will thread that needle better than most people do under pressure.

Reading without reading

Rental agreements, insurance policies, terms of service, long articles — everyone has a pile of documents they know they should read but haven't. Paste the text into an AI chat (or a big chunk of it) and ask: "What are the five things I need to know? Flag anything unusual or that I should be concerned about." You'll get a genuine summary in under a minute.

Important caveat: for anything legally or financially consequential, the AI summary gets you oriented, not off the hook. Use it to know what to read carefully in the original document, not as a replacement for reading it.

Explaining things at exactly the level you need

AI is uniquely patient about re-explaining things in different ways. You can ask for an explanation "like I'm 10 years old," then follow up with "now go deeper" or "explain the technical side." You can ask it to use a specific analogy ("explain this the way you'd explain it to someone who knows cooking but not physics"). This kind of adaptive explanation is something even excellent teachers struggle to do consistently under time pressure — AI does it every time, at no extra cost.

Category 2 — Work

Knowledge workers spend a large fraction of their day on writing, meetings, and communication. These activities are not the core of most jobs — they're the scaffolding around it. AI handles scaffolding well. Here are the most valuable work use cases.

From meeting notes to action items

Every meeting produces some notes. Often they're patchy, unformatted, and scattered across someone's laptop. Paste them into an AI and ask: "Extract the action items. For each, note who owns it and the deadline. If an owner or deadline isn't mentioned, flag it as unassigned." What comes back is a structured task list you can copy into your project management tool immediately.

You can also ask: "What decisions were made?" and "What questions were left open?" These three prompts together turn raw notes into a complete meeting record in about 90 seconds.

Drafting reports, proposals, and job descriptions

The blank page is the hardest part of any writing task. With AI, you never face it. Describe what you need (the document type, the audience, the tone, the key points to cover) and ask for a first draft. The draft will almost certainly need editing — it won't know your company's style, your specific numbers, or the political context of the proposal. But it gives you a structured starting point to react to, which is much faster than starting from scratch.

This is especially useful for document types you rarely write. If you have to write a job description for the first time, or a board-level summary, or a vendor RFP, you may not know the genre conventions. AI does. It will produce a structurally sound template that you fill in with real content.

Analysing data in plain English

You don't have to be a data analyst to ask intelligent questions about a dataset. Paste a table of numbers (copied from a spreadsheet) and ask: "What are the three biggest trends in this data? Which category is growing fastest? Is anything declining?" AI will interpret the numbers and surface patterns that would take a non-analyst much longer to find manually.

This works best with small-to-medium tables (up to a few hundred rows). For larger datasets, you'll need a code-based approach, but for everyday business data — sales figures, survey results, team metrics — conversational data analysis is remarkably effective.

Category 3 — Creative

AI's creative applications get misunderstood in two opposite ways: some people think it's too good ("it's replacing creatives") and others think it's useless ("the output has no soul"). The truth is it's neither. AI is an excellent collaborator for creative work, specifically for the phases of the process where you need quantity before quality.

Brainstorming at scale

Good brainstorming requires generating a lot of options without judging too early. Most people are bad at this alone — they generate three ideas, get stuck, and pick the least-bad one. With AI, you can generate 15, 20, or 30 options in a minute. Your job becomes curation, not generation.

The trick is to give constraints. Not "give me name ideas" but "give me 15 name ideas for a dog-walking app, short and memorable, playful tone, avoid anything with 'paw' or 'wag' since those are taken, mix of single words and two-word phrases." Tight constraints produce better brainstorms. If you like #4 and #11, say so and ask for 10 more in that direction.

Getting unstuck on writing

Every writer knows the feeling: you need a paragraph or a page, you know roughly what it should say, but you can't find the words. AI is genuinely useful here — not to write the thing for you, but to give you options to react to. Ask for three very different versions (lyrical, direct, anecdote-led) and see which one sparks something. You'll often end up discarding all three but knowing clearly what you actually want to write. That's the value: the contrast, not the output.

Editing and tone-shifting

You wrote something. It's okay, but it's too long, or too formal, or too casual. AI is a patient editor. "Make this punchier, cut it by 30%" or "rewrite this in a warmer, more conversational tone" or "make this suitable for a technical audience who already know the domain" — these instructions produce reliable, fast results. You can chain edits: take the AI's version, ask for another pass, and so on until it's close to what you want.

Category 4 — Learning

This is perhaps the most powerful use case for anyone who is serious about learning a new skill or subject. AI removes the two biggest barriers to self-directed learning: access to a patient expert, and the ability to practice interactively.

The adaptive tutor

"You are a patient tutor helping me learn basic economics. I'm a complete beginner. Start with supply and demand — explain it simply, check that I understand, then move on when I say I'm ready." This prompt sets up an interaction that continues at your pace, adapts to what you already know, and never loses patience. You can stay on supply and demand for as long as you need before moving on. No textbook does this.

The key is to push back when something doesn't land: "I still don't understand the bit about equilibrium price — can you try a different analogy?" A human tutor might run out of ways to explain something. AI has essentially unlimited alternative explanations.

Active recall through quizzing

Reading is passive. Being tested is active. Decades of educational psychology research confirm that active recall — trying to retrieve information from memory — is one of the most effective ways to learn. AI makes it trivially easy to quiz yourself on anything. "Ask me 10 multiple-choice questions on the French Revolution. Give me one at a time. After I answer, tell me if I'm right and explain why. Keep track of my score."

This works for any subject: history, science, a programming language, a foreign language, or material from a course you're enrolled in. You can even paste in the content you want to be quizzed on: "Based on this chapter summary, create 8 short-answer questions for me to practice."

Getting feedback on your work

Writing an essay, drafting a business plan, reasoning through a problem — you can get structured feedback from AI on any of these. "Here is my essay draft. Give me honest feedback on the argument, the structure, and the clarity. Don't rewrite it — just tell me where it's strong and where it's weak." This kind of feedback loop, which normally requires access to a mentor or writing group, is available any time for free.

Category 5 — Coding (even for near-beginners)

You do not have to be a developer to benefit from AI coding help. Some of the most valuable use cases are for people who know a little code — enough to understand what a script does and modify it, but not enough to write it from scratch. Here's what that looks like in practice.

Writing small scripts for repetitive tasks

The sweet spot for near-beginner coding with AI is small, well-defined tasks: rename all the files in a folder, combine several CSV files into one, send a formatted email from a spreadsheet, pull data from a website into a table. These tasks take an experienced developer about 20 minutes to write. They take a non-developer potentially never, because the blank page of a code file is impenetrable without a starting point.

With AI: describe the task in plain English, specify your operating system and any constraints, and ask it to explain each line. You get a working script and you understand what it does. If it doesn't work the first time (this happens), paste the error message back in and ask "what's wrong?"

Explaining code you didn't write

You inherited a script. Or you found something on the internet that almost does what you need. Or you wrote something yourself six months ago and can't remember how it works. Paste it into AI and ask: "Explain what this code does. Go line by line for anything complicated. Assume I know Python basics but not advanced concepts." What comes back is a plain-English walkthrough of every significant part of the code.

Debugging

Error messages are designed for compilers, not humans. They're often cryptic, referencing line numbers and jargon that means nothing without context. AI translates them. Paste the error message, paste the relevant code, and ask: "What's causing this error and how do I fix it?" You'll get both the fix and an explanation of why it was broken — which means you'll understand the concept for next time.

Why the "before" matters as much as the "after"

One thing the prompts above share: they all start with something real that you already have. Meeting notes, raw thoughts, a block of code, a chapter of a textbook. This is intentional. AI doesn't create from nothing — it transforms what you give it. The better and more specific your input, the more useful the output.

The before/after gallery below shows exactly what this looks like in practice: messy raw input goes in, and polished, usable output comes out. More importantly, it shows the prompt that made the transformation happen.

Recognising which category a task belongs to

After working through enough examples, you start to develop a mental reflex: you see a task and immediately know the shape of the prompt that will handle it. Before that reflex kicks in naturally, it helps to be deliberate about categorisation.

Each of the five categories has a characteristic "prompt shape":

• Everyday life tasks tend to involve transforming something messy into something polished, or explaining something to yourself. Shape: "take this [messy input] and give me [clean output]."
• Work tasks tend to involve structuring information, drafting documents, or extracting insights. Shape: "here is raw material, turn it into [structured output] for [audience]."
• Creative tasks tend to involve generating options to react to. Shape: "give me N options for [creative thing], I'll choose and iterate."
• Learning tasks tend to involve dialogue: explanation, testing, feedback. Shape: "explain [thing] to me at [level], then check my understanding."
• Coding tasks tend to involve writing, explaining, or fixing code with a clear success condition. Shape: "write/explain/fix [code thing], explain why."

One last thing: start with something real

Everything in this lesson becomes real skill the moment you use it on an actual task from your life. Not a practice exercise, not a hypothetical — a real email you've been avoiding, a document that's been sitting in your inbox, a piece of writing you're stuck on.

The thing that separates people who get a lot of value from AI and people who feel like it "never works for me" is almost always the same: the first group uses it on real tasks with real stakes. The second group tries it on abstract tests and gets abstract results.