What AI Can Do For You

30 min beginner Lesson 1

Learning Outcomes

  • Describe the five core things an AI assistant does well for office work
  • Recognise tasks in your own week that AI can speed up
  • Compare a task done with and without AI to judge the real time saved
  • Identify where AI fits as a first-draft partner, not a final authority
  • Spot the limits of AI so you know what to double-check

Lesson Plan

Segment Duration Topic
Intro 3 min What this lesson covers and what an AI assistant is
Capability 1 5 min Summarising long documents and threads
Capability 2 5 min Drafting and rewriting text
Capability 3 4 min Answering questions and explaining
Capability 4 4 min Making sense of data
Before & after 4 min A real task with and without AI
Where it fits 3 min Treating AI as a first-draft partner
Wrap-up 2 min Key takeaways and next lesson

Before You Begin

Pre-work:

  • Nothing technical — no installs, no command line. If you can open a web browser and a document, you are ready.
  • Skim the Practical AI course landing page so you know where this lesson sits.
  • Have one real piece of your own work nearby — an email thread, a report, or a spreadsheet — for the exercises.

Shopping List:

  • A web browser (the one you already use is fine)
  • A free account with any mainstream AI assistant — we compare the options in Lesson 2: Choosing the Right AI Tool
  • Fifteen quiet minutes to try the prompts

A note on terms: an AI assistant (a "chatbot" or "large language model") is a tool you talk to in plain English through a website. You type a request, it writes back — the whole interface.

1 Capability One: Summarising Long Things

The single most useful thing AI does is turn long content into short content. A forty-minute transcript becomes six bullets; a twenty-page policy becomes a one-paragraph brief; a messy thread becomes a clear list of decisions and open questions. You give it the text, tell it the shape you want, and it produces it in seconds.

> Summarise the report below in five bullet points an executive could act on. Flag anything needing a decision. [paste the report]

You control the output's shape:

You ask for You get
"Summarise in 3 bullets" A skimmable overview
"List every action item with its owner" A task list from the noise
"What are the open questions?" Just the unresolved items
TIP
Tip
Always tell the AI the format you want — bullets, a table, a paragraph, a length. The more you describe the output, the less editing you do afterwards. Five bullets beats summarise this.
WARNING
Watch Out
A summary is only as good as the source you paste in — give it half the thread and you get half the picture. Check it did not drop something.
2 Capability Two: Drafting and Rewriting Text

The blank page is where hours disappear. AI removes it: you describe what to write, and it hands you a draft to react to — far faster than producing one from nothing. This works for almost any text at work: a stakeholder email that delivers bad news gently, a status update from rough notes, or a polite decline to a meeting invite.

> Draft a short, warm email telling a client the delivery date has slipped two weeks due to a supplier delay. Reassure them, keep it under 120 words, and end with a clear next step.

It also rewrites text you already have — paste a paragraph and ask for a different tone, a shorter version, or a fix for clumsy wording.

> Rewrite the paragraph below to sound more confident and less apologetic, same length. [paste your paragraph]
NOTE
Key Insight
AI is a first-draft engine, not a final-word authority. Sending an unedited AI draft is like forwarding the first thing an intern hands you — never bet your name on it unread.
3 Capability Three: Answering Questions and Explaining

You can ask an AI assistant to explain anything in language that fits you. Unlike a search engine, it answers your question directly instead of handing you ten links. That is useful for understanding a term in a contract, getting the gist of a topic before a meeting, or turning a complicated idea into something you can explain to your team.

> Explain what "working capital" means in plain English, with an example a non-finance manager would understand. Then give me three questions to ask our finance team.

You can also push for the right level of depth:

Ask this Get this
"Explain it like I'm new to the topic" A gentle, jargon-free overview
"Give me the executive version" The bottom line in two sentences
"What am I likely to be missing?" Blind spots and follow-ups
WARNING
Watch Out
AI can sound completely confident while being wrong — a hallucination. For facts, figures, names, and dates, treat the answer as a lead to verify, not a fact to quote. Lesson 8 covers verification.
4 Capability Four: Making Sense of Data

You do not need to be a spreadsheet expert to get answers out of numbers. Paste a small table and ask plain-English questions — or have AI write the formula you half-remember.

> Here is last quarter's sales by region. Which grew fastest, which shrank, and what stands out? Then write an Excel formula that sums column C only where column A says "Approved", and explain each part. [paste the table]

What it does well versus where to be careful:

Good for Be careful with
Spotting trends in data you paste in Precise arithmetic on long lists
Explaining what a chart or table shows Numbers it was not given
Drafting and explaining formulas Filing-grade figures, unchecked
TIP
Tip
Where the exact number matters, let AI find the trend or write the formula, but do the final sum in your spreadsheet. AI is great at the so what, less reliable at the exactly how much. Lesson 5 digs into data work.
5 A Before-and-After: The Monday Status Report

Capabilities are abstract, so let us walk one through.

Without AI. It is Monday. You open last week's report, copy the structure, then scroll back through channels and emails to remember what happened. You write each section from scratch, second-guess the tone, and forty-five minutes later send something you are not thrilled with.

With AI. You paste your rough notes from the week and ask for the report:

> Turn these rough notes into a weekly stakeholder status update with sections Progress, Risks, Next Steps. Under 200 words, professional but not stiff. [paste your notes]

You get a structured draft in seconds, then spend ten minutes correcting two facts, softening a sentence, and adding a detail only you knew. You send something you are happy with — twelve minutes against the usual forty-five.

The point is not that AI wrote your report. You still decided what mattered, fixed what was wrong, and put your name on it. AI removed the slow part — the blank template and wrestling words into shape — so your time went to judgement, not typing.

TIP
Tip
The biggest wins are recurring tasks you find tedious: weekly reports, meeting notes, routine replies. Build a prompt that works once, then reuse it. Lesson 9 turns these into reusable templates.
6 Where AI Fits — and Where It Does Not

The healthiest mental model: AI is a fast, tireless junior colleague who is occasionally confidently wrong. It is fantastic for first drafts, summaries, explanations, and brainstorming, but should never be the final signature on anything carrying your judgement. A sorting test:

Great fit for AI Keep human-led
First drafts of emails, reports, proposals Final sign-off and the decision itself
Summarising long documents and threads High-stakes legal or financial commitments
Brainstorming and explaining topics Confidential data and anything expensive to get wrong
Reformatting and tidying existing text Personal, sensitive, or HR-delicate messages

The skill you build here is not "using AI" — it is knowing which column a task belongs in. Used well, AI does not replace your expertise; it clears the busywork that buries it.

WARNING
Watch Out
Never paste anything into a public AI tool that you would not send outside your organisation — customer records, salaries, unreleased financials, anything confidential. Where your data goes depends on the tool and plan. Lesson 10 covers this.

Questions & Answers

Q: Will AI replace my job?
It is far more likely to change the shape of your work than replace it. AI is strong at the mechanical parts and weak at judgement, accountability, and relationships — the parts your job rewards. The realistic risk is not "AI takes your job" but "a colleague who uses AI well does more in the same hours." This course puts you in that group.
Q: I am not technical at all. Is this for me?
Yes — and that is the point. There is nothing to install and no code: you type a request in plain English on a website and read what comes back. If you can write an email, you already have every skill the interface needs. The rest is practice.
Q: How do I trust an answer if AI can be wrong?
You do not trust it blindly — you treat it like a capable but new colleague. Lean on it for drafts, structure, and explanations, and verify anything factual before it leaves your hands: names, numbers, dates, quotes. Use AI where a small error is easy to catch; double-check where it is expensive.
Q: Is it cheating to use AI for my work?
No more than using a spreadsheet instead of a paper ledger. What matters is that the output reflects your judgement and meets your organisation's policies. If your workplace has rules about disclosing AI use, follow them — Lesson 10 covers this. Using AI to think faster is fine; using it to skip thinking is not.

Key Takeaways

  1. Five core capabilities — AI summarises, drafts and rewrites, answers and explains, makes sense of data, and brainstorms. Most office use cases mix these.
  2. AI is a first-draft partner — It gets you past the blank page; you handle judgement, errors, and ownership.
  3. Describe the output you want — Format, length, tone, and audience in the request mean far less editing later.
  4. The real win is recurring busywork — Weekly reports, meeting notes, and routine replies are where a reusable prompt pays off.
  5. Verify what matters — AI can sound certain while wrong, so double-check facts, figures, and anything expensive first.
  6. Know which column a task belongs in — Sort AI-friendly work from human-led work, and never paste confidential data into a public tool.

Next Steps: Lesson 2: Choosing the Right AI Tool