10 tasks ChatGPT solves for you today

Here's a road-tested map of the jobs where ChatGPT saves you time on day one. Each comes with a ready phrasing — drop your own details in.

Writing

Let's start with what people open an AI for most often — the texts you have to write and don't want to.

1. Emails and messages. "Write a polite rejection to a supplier: we went with someone else, keep the way back open."
2. Fixing your own text. "Rewrite this clearer and half as long: [text]."
3. Posts and descriptions. "3 versions of a post about [event] for the company's social media, lively tone."

Thinking

Next comes the part that surprises people most: an AI helps not only when you have to write, but when you have to think.

4. Explaining hard things. "Explain [topic] to someone with no technical background, with an analogy."
5. Brainstorming. "20 gift ideas for a colleague who's into mountains and coffee. Budget up to 50 €."
6. An outside view. "Here's my plan. Find the weak spots and ask the questions I'm missing."

Information

The third group is about other people's texts: long, dull, and written so that nobody reads them.

7. Making sense of documents. Paste the contract text: "List my obligations and the risks in this contract."
8. Summarising. "Give me this article in 5 bullet points: [text]."
9. Tables and structure. "Turn these notes into a table: task, deadline, owner."
10. Learning. "Build me a month-long plan to learn [skill] at 30 minutes a day, with practice."

Remember: the AI can invent a fact, a number or a link — they sound convincing, but they don't exist. Check them — especially in documents and anything legal.

What if you have no job, or not an office one

The list above reads as though the reader has an employer, reports and clients. Plenty of people don't, and it doesn't matter: the same ten jobs work in a different life, they're just called something else.

• Looking for work. A CV rewritten in human language. A cover letter for one specific vacancy rather than one for all of them. Interview practice: «you're the interviewer, ask questions one at a time and then go through my answers».
• Running your own thing. Product descriptions in batches from one template, replies to reviews, the text of an ad, working through a supplier contract.
• Home and admin. A letter to the building management — polite but firm. An unexplained line on a bill. Medicine instructions in tiny print. A trip plan with children and a budget.
• Studying. Explaining a topic three different ways until it lands. A revision plan. Checking your own text for weak spots before a tutor sees it.

The jobs are the same — letters, analysis, structure, explanations. Only who they're addressed to changes.

Where the time actually goes

The saving doesn't come from magic but from one shift: the draft stopped being your job. The slowest part used to be the blank page — the first three paragraphs of an email, the first shape of a table, the first attempt at a summary. Now the draft appears immediately and your work starts at editing, which goes considerably faster than writing from nothing.

How many hours that is, measure on your own tasks: time how long an email takes you now and how long it takes with a draft ready. Other people's savings figures are useless here, because everyone's emails are different.

Getting the most out of these phrasings

Two rules turn templates into results:

1. Add your context. Every phrasing gets stronger with detail: not just "write a rejection to a supplier", but "…we've worked together two years, we're parting over price, we want to keep the relationship". Remember the rule from the first-dialogue lesson: situation + constraints + result.
2. Don't settle for the first answer. You have a draft — now go with "shorter", "different tone", "add numbers". The second pass is almost always noticeably better.

What's missing from the list, and why

There's no "find me market statistics" and no "diagnose this". The first, because a language model answers from what it learned and can invent a convincing-looking number — that's called a hallucination, and it sounds exactly as confident as the truth: for facts with sources there are web-search modes and dedicated tools. The second, because health, money and law: AI is great at "explain the terms in these test results", but a professional makes the call. Knowing the boundaries is as much a skill as phrasing a good request.

Task 11, the one everyone forgets

Rehearsal. The AI is a perfect sparring partner: "You're an HR manager, interview me for a project manager role, one question at a time, and review my answers at the end." People rehearse salary negotiations, thesis defences and awkward landlord conversations the same way. It's the job people remember last — and it's a training partner that never tires and never judges.

The first-week method

Don't try to adopt all ten at once — that's the surest way to adopt none. Simpler plan: today, pick one job and run it on something real. Tomorrow, the next one. Over a week you'll work through several, and some will stick and go daily. That's normal: everyone's top list is different. An accountant keeps the tables and contracts, a teacher keeps the explanations and lesson plans, a salesperson keeps the emails and objection rehearsals.

The combo: chaining jobs together

The real savings start when jobs link into a pipeline inside one chat. Example: paste a long report → "summarise in 5 bullets" (job 8) → "turn the bullets into a table: finding, number, risk" (job 9) → "write a short email to the director based on the table" (job 1). A few minutes, and a raw document has become a summary, a table and an email. One source, one conversation, three finished things.

Do this now

Pick the one job out of the ten that you actually need today, drop your context in and send it. Then ask, in one sentence, to improve the result. That's the basic AI loop: draft → edit → done.