You can
just ask.

AI tools like Claude and ChatGPT can now write, research, analyze, plan, and build things. The whole interface is a box you type into. Most of us still use it like a chattier search engine.

You can just ask it for the thing.

Message

Try this one in Claude ChatGPT

Free to try.
No coding.
Mostly typing.

Things you can just ask for

Every one of these works today, in the free versions of Claude or ChatGPT unless it says otherwise. Steal them. Swap in your own details. Hit a button and it opens with the ask already typed.

    Square brackets like [paste it here] are for your stuff. Replace them before you hit send.

    The ask ladder

    More than a billion people use ChatGPT every week, and most conversations are about getting information, advice, or writing help. That’s the bottom of the ladder. The view gets very good from about rung three.

    1. 1

      Ask it questions

      Like search, except it answers in sentences and you can say “wait, what does that mean?”

      “Why is my sourdough flat?”

    2. 2

      Give it your stuff

      Paste the email. Upload the PDF. Snap a photo of the whiteboard. Answers about your situation beat answers about everyone’s.

      “Here’s the 60-page report. What does it say about volunteers?”

    3. 3

      Have it make things

      Drafts, tables, charts, slide outlines, flyers, spreadsheets, working mini-apps. You give notes, like an editor.

      “Make it a table. Now a chart. Now make it less ugly.”

    4. 4

      Think with it

      Have it interview you, poke holes in your plan, play the skeptical funder, or walk you through a decision.

      “Ask me questions one at a time until you get it.”

    5. 5

      Let it do things

      Connect your email, calendar, and docs. Newer agent features can browse websites, work with files on your computer, and handle multi-step tasks. More power means more care, so start with drafts you review.

      “Find the three emails I forgot to answer and draft replies.”

    Past rung five, there’s a deep end ↓

    The deep end

    The strongest models right now (as of late 2026, Claude Opus 5.5 and OpenAI’s GPT-6 family) don’t just answer. Paired with agent tools like Claude Code, Claude Cowork, Codex, and ChatGPT Work, they can work on their own for minutes or hours: writing code, running it, checking their own work, and fixing what broke.

    None of these need you to code. They do need a paid plan, patience, clear notes, and a human checking the result.

      Don’t know what to ask?
      Ask it what to ask.

      This is the move almost nobody makes. Tell it about your work and let it interview you. Fill in the blanks, and we’ll write the ask for you.

      I work as . Every week I spend way too long on . The part I secretly dread is .

      Your ask

      Ask Claude Ask ChatGPT

      How to ask (it’s not a magic spell)

      You don’t need “prompt engineering.” You need the skill you already use on humans: explaining what you want.

      • Treat it like a brilliant new coworker

        Smart, fast, tireless, and has never met you. It doesn’t know your org, your audience, or that Dave is touchy about the budget. Tell it.

      • Paste the actual thing

        Don’t describe the email. Paste the email. Don’t summarize the spreadsheet. Upload the spreadsheet.

      • Say who it’s for

        “For a busy board member.” “For my 9-year-old.” “For a skeptical journalist.” The same facts land very differently.

      • Let it ask you questions

        End with “Ask me anything you need before you start.” It usually asks the questions you forgot to answer.

      • The first answer is a draft

        Push back. “Shorter.” “Less corporate.” “That’s wrong, here’s why.” “Give me five more.” It doesn’t get offended. It gets better.

      • Ask for options

        “Give me ten subject lines” beats “give me a subject line.” Then pick, or say “mix 3 and 7.”

      Okay, but…

      Reasonable questions, honest answers.

      Doesn’t it make stuff up?

      Yes, sometimes, and with a straight face. It’s great at drafting, explaining, organizing, and brainstorming. It’s less reliable on exact facts, numbers, quotes, citations, and very recent events.

      Treat it like a sharp intern: trust the first draft, verify the facts that matter. Ask it to show sources, then click them. For medical, legal, or financial calls, use it to prepare your questions for a professional rather than to replace one.

      Claude or ChatGPT? Which one?

      Either. Truly. They’re both excellent, both have free versions, and they leapfrog each other every few months. Try the same ask in both and keep the one whose answers you like. Lots of people use both.

      Paid plans start around $20 a month (ChatGPT also has a cheaper Go plan) and give you more usage, the strongest models, and extras like deep research. Start free. Upgrade when you keep hitting the limits.

      Work at a nonprofit? Both companies offer discounts. Claude for Nonprofits offers its Team plan at a deep discount (even cheaper in low- and middle-income countries), and OpenAI for Nonprofits offers up to 75% off ChatGPT Business.

      Is it safe to paste my work stuff in?

      It depends on your plan, your settings, and your organization’s rules, so check those first. A few sensible defaults:

      • Don’t paste passwords, ID numbers, or other people’s personal information (donor lists, client records) unless your organization has approved a tool for that.
      • Decide whether your chats can help train future models. In Claude it’s Settings → Privacy → “Help improve our AI models.” In ChatGPT it’s Settings → Data controls → “Improve the model for everyone.” Switch it off if you’d rather they didn’t.
      • Team and business plans generally come with stronger privacy terms. If you’re using it for work, ask whoever runs your IT.
      Isn’t AI bad for the planet?

      Training giant models and building data centers uses a lot of energy, and that deserves real scrutiny. A single typical text question uses less than most people guess, though. Google says its median text prompt uses about 0.24 watt-hours, like watching TV for under nine seconds. OpenAI’s CEO says an average ChatGPT query uses about 0.34 watt-hours, roughly an oven running for one second. Those are the companies’ own numbers, so hold them loosely.

      Long reasoning and agent tasks use a lot more, and so does generating video. Use the heavy stuff on purpose, and the everyday stuff without guilt.

      I tried it once and it was… meh.

      Two likely reasons. First, it probably had too little to go on. Try again with the real document pasted in and three sentences about who it’s for. Second, the tools change fast. The version you tried a year or two ago is not the one you’d use today.

      Isn’t this kind of cheating?

      Is a calculator cheating? A spellchecker? An assistant? You’re still the one who decides what’s good, what’s true, and what gets sent. You own the result. Just be upfront where it matters, like school assignments or anywhere people expect your own words.

      Do I need to learn to code?

      Nope. You can ask it to build a calculator, a quiz, a chart, or a tiny game and it will just… do that, in the chat. If you want to learn to code, it’s also the most patient tutor you’ll ever have.

      Your homework:
      ask it one thing today.

      Something you’ve been putting off. Something mildly annoying. Something you assumed a computer couldn’t do.