Your learning path, one lesson at a time.

The idea: clear instructions (a spec) - helpers do exactly what you say, not what you meant · Build your own

Do: Write step-by-step instructions for making a sandwich. A family member follows them EXACTLY as written - no guessing. Fix your steps and try again.

Copy this prompt:

With an adult, on their account: "I'm 11 and I wrote instructions for making a sandwich. Don't rewrite them. Ask me two questions that show me where a robot would get confused."

You'll know it worked when: Someone followed your instructions start to finish without asking you anything.

Think about: Which step did you think was obvious, but wasn't?

The idea: breaking a big goal into small tasks - how every builder and every AI team starts · How evolution works

Do: Pick a real job at home or school (a birthday, a garden, a class event). Write the goal at the top, then break it into at least 8 small steps that each take under 15 minutes.

Copy this prompt:

With an adult: "Here is my plan for a class event. Don't add steps. Ask me which step I'd do first, and why."

You'll know it worked when: Every step is small enough that you know exactly how to start it.

Think about: Which step will be hardest, and who could help?

The idea: verifying - good builders check their helpers' work · Build your own

Do: With an adult, ask an AI helper a question about something you already know well (a sport, an animal, a game). Find one thing it says, and check it in a book or a second trusted source.

Copy this prompt:

"Tell me three facts about [something I know well]. Then tell me which one you're least sure about."

You'll know it worked when: You found one claim and checked it yourself - true, false or not sure.

Think about: How would you know if a helper was wrong about something you DON'T know well?

The idea: building something small and getting feedback before it's perfect · Build your own

Do: Draw an app or a game on paper, one screen per page. Show it to someone and let them 'tap' with a finger while you flip the pages. Change one thing they found confusing.

Copy this prompt:

Optional, with an adult: "I drew a paper app for [idea]. Ask me three questions a first-time user might have."

You'll know it worked when: Someone used your paper app, and you changed one thing because of them.

Think about: What surprised you about how they used it?

The idea: directing a helper - you set the rules for how it helps you · Build your own

Do: Pick one homework topic that's hard for you. Use the tutor prompt and work through it - no answers allowed until you've tried.

Copy this prompt:

"Be my tutor for [topic]. Don't give me the answer. Ask me one question at a time, wait for my reply, and give me a hint only if I'm stuck twice."

You'll know it worked when: You reached the answer yourself, and you could explain each step.

Think about: When did you most want to just ask for the answer - and what happened when you didn't?

The idea: explaining is how you find what you don't understand yet · Build your own

Do: Explain today's lesson to the AI as if it were a younger student. Let it ask you questions.

Copy this prompt:

"I'm going to explain [topic] to you like you're 10. Ask me the questions a curious 10-year-old would ask, one at a time. Tell me what I left out at the end."

You'll know it worked when: You answered at least three questions without looking anything up.

Think about: What's the one part you'd explain better next time?

The idea: reviewing a helper's work - the most important builder skill · Build your own

Do: Ask an AI to explain something you're good at. Find anything wrong, missing or too confident, and check one claim with a second source.

Copy this prompt:

"Explain [something I'm good at] in five sentences. Then rate how sure you are of each sentence."

You'll know it worked when: You found and checked at least one claim yourself.

Think about: Was the AI's confidence a good guide to whether it was right?

The idea: measuring your own progress - builders improve what they track · How evolution works

Do: Once a week, write three lines: what I learned, where I struggled, what I'll try next. Keep it private or share it with a parent or teacher.

Copy this prompt:

"Here are my three lines from this week. Ask me one question that helps me choose next week's goal. Don't choose it for me."

You'll know it worked when: You have two weeks of entries.

Think about: What do you notice when you read last week's entry?

The idea: the gap - which kinds of work and which places are changing fastest, and why · Why the world is changing

Do: Open our Road to Orbit and Concerns pages. Pick ONE change from this week that touches your town or your family (for example, new data centers needing power and water, or jobs that are changing). Ask one adult how it affects them.

Copy this prompt:

"I read that [the change]. Don't tell me what to think. Ask me three questions that help me work out who it helps, who it leaves out, and what a young builder could do about it."

You'll know it worked when: You can explain one real, dated change in your own words - who it helps and who it leaves out.

Think about: What could a builder your age do about it - even something small?

The idea: a builder's first budget is time - plan it honestly, keep your promises, and keep room for rest, friends and life off screen · The builder economy

Do: On paper, plan one week: school, family, sleep, friends, time outside, and then the time you have left for building. Make one small building promise you can keep this week. At the end of the week, check: did you keep it, and did you still rest?

Copy this prompt:

"Here's my week plan and my one building promise. Don't change it. Ask me two questions about whether it's realistic, and one about whether I left enough time for rest and people."

You'll know it worked when: You kept your promise - or you can say honestly why not and what you'll change next week.

Think about: What did you protect in your week that matters more than building?

The idea: career paths - college, the trades and technical work that build the world's power, factories and data centers, and building your own; every road is real, and you can combine them · Career paths

Do: On paper, draw three roads: college, the trades and technical work, and building your own. Under each, write what you already know - one job, one thing you'd learn, one thing you'd like and one you wouldn't. Then look up one real job on each road and add what surprised you.

Copy this prompt:

"Here's my map of three career roads and what I wrote under each. Don't tell me which road to pick. Ask me two questions about each road that help me see it from all sides - what's exciting, what's hard, and what it asks of me."

You'll know it worked when: A three-road map with at least one real job on each road, and one question you still want answered.

Think about: Which road surprised you most - and is there a way to combine two?

The idea: learning from people already on the road - the best career advice comes from real work, not guesses · Career paths

Do: With a parent or teacher's help, pick one adult you already know whose work interests you (a relative, a neighbor, a family friend, a school visitor). Write five questions first - what a normal day is like, how they learned, what AI is changing in their work, what they'd tell their younger self. Talk to them, with your adult arranging it, in person or on a family call.

Copy this prompt:

"I'm going to ask an adult I know about their work. Here are my five questions. Don't write new ones. Tell me which of mine might get a short answer, and ask me how I could make it more open."

You'll know it worked when: You asked your five questions and wrote down the one answer that surprised you most.

Think about: Did their road look like you expected - and what would you ask next time?

The idea: builders start from people, not from tools · Why the world is changing

Do: Ask two people you know (family, a coach, a shop owner) what wastes their time each week. Write down their exact words.

Copy this prompt:

"Here's what two people told me wastes their time. Don't suggest solutions yet. Ask me questions that help me understand the real problem."

You'll know it worked when: You can say the problem in one sentence, in the person's own words.

Think about: Which problem do YOU care about solving, and why?

The idea: the spec - goal, who it's for, limits, and how you'll know it worked · Build your own

Do: Write half a page: the goal, who it's for, what it must NOT do, and the test that shows it works.

Copy this prompt:

"Here's my spec. Don't rewrite it. Point out anything unclear and ask me how I'd test each part."

You'll know it worked when: Someone else can read your spec and tell you what you're building.

Think about: What did writing the test teach you about the goal?

The idea: orchestrating a helper step by step - you decide every step · Build your own

Do: Build the smallest version that could help: a simple page, a spreadsheet, a checklist, a schedule. Ask the AI for one step at a time and run each step before the next.

Copy this prompt:

"I'm building [my spec]. Give me only the next small step and explain why. After I try it, I'll tell you what happened."

You'll know it worked when: Version 1 works on your own test from the spec.

Think about: Where did you overrule the AI - and were you right?

The idea: feedback loops - real use beats your own opinion · How evolution works

Do: Watch one person use it without helping them. Write down where they got stuck. Change one thing.

Copy this prompt:

"In general terms - no names or personal details - here's where my tester got stuck. Ask me which change would help them most, and why."

You'll know it worked when: Version 2 fixes the place your tester got stuck.

Think about: What did the tester do that you never expected?

The idea: teaching is how builders grow - and how communities grow · The builder economy

Do: Show a friend or a class how you built it, step by step, including what went wrong.

Copy this prompt:

"Help me plan a 3-minute talk about how I built [thing]. Ask me questions; don't write it for me."

You'll know it worked when: Someone else could start building something similar after your talk.

Think about: What would you build next?

The idea: the new builder economy - one person with AI helpers can now offer a useful service · The builder economy

Do: For your project, write who it helps, what it saves them (time, money, worry), and what a fair way to support it would be - free, a small fee, a sponsor, or a volunteer project. No real money changes hands in this lesson.

Copy this prompt:

"Here's who my project helps and what it saves them. Ask me questions about whether it's worth offering to more people - don't give me a business plan."

You'll know it worked when: You can say in two sentences who it helps and how it could keep going.

Think about: What would make it useful enough that people ask for it?

The idea: responsibility - every project reaches more people than its first user · Global responsibility

Do: For your project, list everyone it could touch: the person it helps, their family, people who don't use it, people in other places, and the planet (the electricity and devices it needs). Write one way it could help and one way it could cause a problem for each.

Copy this prompt:

"Here's my list of who my project touches. Don't add to it. Ask me about one person or place I forgot, and how they might see it."

You'll know it worked when: Your list has at least four groups, each with a help and a possible problem you thought of yourself.

Think about: What is one change you could make so your project is fairer to someone you forgot at first?

The idea: every real project has a budget - time, tools and money - and a great one can start at zero · The builder economy

Do: Write your project's budget in three lines: the hours it will take, the free tools you'll use, and the money it needs (aim for $0). Then add one 'someday, maybe' line: a tool that could help later, what it would need to do to be worth it, and who would decide. This is a plan, not a purchase - nobody buys anything.

Copy this prompt:

"Here's my project budget. Don't add costs. Ask me whether a free tool could do each job, and one question about whether my 'someday' tool would really be worth it."

You'll know it worked when: Your budget fits on a card, uses only free tools, and your 'someday' line says what would make it worth it.

Think about: What did your project need more of - money, or time and good ideas?

The idea: honesty and credit - say what helped you (people, sources, AI) and be truthful about what your project can't do · The builder economy

Do: Write a short 'how I built this' note for your project: who helped, which sources you used, what an AI helper did and what you did yourself, and one thing it can't do yet. Show it to the person you built it for.

Copy this prompt:

"Here's my 'how I built this' note. Don't rewrite it. Ask me whether I've left anyone or anything uncredited, and whether I've promised more than it can do."

You'll know it worked when: Your note credits every helper, says honestly what the AI did, and names one limit.

Think about: How did it feel to give credit away?

The idea: two roles - one helper builds, another checks, and you decide · Build your own

Do: Use one AI chat as your builder and a second, separate chat as your reviewer. Paste the builder's work into the reviewer and ask for problems. You choose what to fix.

Copy this prompt:

Reviewer: "You're a careful reviewer. List the three biggest problems in this work, most important first. Don't fix them."

You'll know it worked when: You fixed at least one problem the reviewer found - and rejected one you disagreed with.

Think about: Why did you reject the one you rejected?

The idea: goals and thresholds - you can't improve what you don't measure · How evolution works

Do: For your project, write one goal and one number that shows progress (people who used it, minutes saved, mistakes caught). Set a target for next week.

Copy this prompt:

"Here's my goal and my measure. Ask me whether this number really shows the goal is met, and what could fool it."

You'll know it worked when: You have a number for this week and a target for next week.

Think about: What would make your measure look good while the goal isn't met?

The idea: iterate - plan, act, measure, improve, every week · How evolution works

Do: Run one full week: plan on Monday, build during the week, measure on Friday, write one improvement for next week.

Copy this prompt:

"Here's my week: plan, what I did, my measure. Ask me two questions that help me pick ONE improvement."

You'll know it worked when: You finished one full loop and wrote your one improvement.

Think about: What changed between Monday's plan and Friday's result?

The idea: human judgment - honesty, privacy, credit and care stay with you · Build your own

Do: List the decisions in your project that a helper should never make for you: what's fair, what's private, who gets credit, who could be hurt.

Copy this prompt:

"Here are the decisions I keep for myself in my project. Ask me about one I might have missed."

You'll know it worked when: Your list has at least four decisions, and you can say why each one is yours.

Think about: Which decision was hardest to hand to no one but yourself?

The idea: every builder who teaches makes two more builders · The builder economy

Do: Teach a friend the builder-and-reviewer method on their own small project.

Copy this prompt:

"Help me plan how to teach a friend the builder-and-reviewer method in 20 minutes. Ask me what they already know first."

You'll know it worked when: Your friend ran one builder-and-reviewer round on their own.

Think about: What did teaching it show you about your own method?

The idea: global responsibility - thinking from all sides about who your work affects, near and far · Global responsibility

Do: Pick three very different people your project or idea could affect - someone your age in another country, someone much older, someone who can't afford a device. Write how each of them would see it, in their words, before asking any AI.

Copy this prompt:

"Here's how I think three different people would see my project. Don't tell me what they'd think. Ask me two questions that show me a view I might have missed."

You'll know it worked when: You changed or added one thing in your project because of a view you hadn't thought of.

Think about: Whose view was hardest for you to imagine, and why?

The idea: long-horizon planning - good builders think past next week to the world twenty years from now · Global responsibility

Do: Imagine your project, or the problem it solves, in 2046. Write what would have to be true for it to still be helping people then, and what could go wrong over twenty years (people, places, the planet). Pick one thing you can do this month that makes the good future more likely.

Copy this prompt:

"Here's my 2046 picture for my project. Don't fill it in. Ask me one question about a risk I haven't planned for and one about who should have a say."

You'll know it worked when: A short plan with one long-range risk and one action you take this month.

Think about: What do you want the world to be like for people who are ten years old in 2046?

The idea: riding the wave - you don't need the whole career planned, only the first real rung you can climb this year · Career paths

Do: For your favourite road (or two), find one first rung you could start this year at $0: a free course, a club, a library or community program, a school class, a summer job or volunteer role near the build-out, or a tiny service of your own. Write the rung, the first step, and the date you'll take it.

Copy this prompt:

"Here is the first rung I picked and my first step. Don't pick for me. Ask me two questions about whether it's really free and really this year, and one about who could help me start."

You'll know it worked when: One written first rung, a first step, and a date - and nothing on it costs money.

Think about: Who could you bring with you onto this rung?

The idea: thinking before asking - the guess is where the learning happens · Build your own

Do: Pick one question the whole family wonders about (why the sky is blue, how bread rises, where your town's name comes from). Everyone writes or says their own guess first - the youngest goes first. Then one adult asks the AI helper, and you compare.

Copy this prompt:

"Our family has guessed the answer to [question]. Here are our guesses. Don't tell us the answer yet. Tell us which guesses are closest, and give us one hint so we can try again."

You'll know it worked when: Everyone made a guess before the AI answered, and the family found the answer together.

Think about: Whose guess was closest - and was it who you expected?

The idea: verifying - AI can sound sure and still be wrong, and every age can catch it · Build your own

Do: Ask the AI helper about something someone in the family knows really well (a grandparent's old job, a child's favorite animal, a family recipe). The family expert listens and calls out anything wrong. Check one claim in a book or a second trusted source.

Copy this prompt:

"Tell us five facts about [something our family knows well]. Then tell us which one you're least sure of."

You'll know it worked when: Your family expert found a mistake, or checked every fact and found none - and one claim was checked in a second source.

Think about: How did it feel to know more than the AI about something?

The idea: agreeing the rules together - every generation gets a say · How evolution works

Do: Everyone suggests one rule (what AI may help with, what we always do ourselves, what we never share, when screens go off). Vote, keep the best three to five, and write them on paper for the fridge. Revisit in a month.

Copy this prompt:

"Here are our family's rules for using AI. Don't add new rules. Ask us about one situation our rules don't cover yet."

You'll know it worked when: A written agreement with a rule from at least two generations, and a date to revisit it.

Think about: Which rule came from the person you least expected?

The idea: the gap - work and daily life change across generations, and each one has lived a part of it · Why the world is changing

Do: The oldest person at the table tells how one job or everyday task worked when they were young (a phone call, a letter, a shop, a factory). The youngest says how it works now. Together, guess how it might work when the youngest is grown up.

Copy this prompt:

"Our family is talking about how [a job or everyday task] changed from [then] to now. Don't predict the future for us. Ask us two questions about what changed, who it helped and who it left behind."

You'll know it worked when: You wrote down one 'then', one 'now' and one family guess about 'next'.

Think about: What has stayed the same through all the changes?

The idea: good projects start with a real problem for real people · Build your own

Do: Everyone names one small annoyance at home (forgotten chores, lost recipes, a messy shed, a plant nobody waters). Pick one the family agrees on. Write it as one sentence: who has the problem and what would make it better.

Copy this prompt:

"Our family wants to solve this problem: [one sentence]. Don't solve it for us. Ask us three questions that help us understand the problem better."

You'll know it worked when: One problem, written in one sentence, that everyone agreed on.

Think about: Why did the family choose this one?

The idea: a team splits the work - like a team of AI agents, each with a clear role · How evolution works

Do: Break your project into small steps on paper. Give everyone a job that fits them: someone draws, someone writes, someone tests, someone checks. Set one small promise each for the week.

Copy this prompt:

"Here are our project steps and who does what. Don't change our plan. Ask us which step might be hardest and who could help with it."

You'll know it worked when: Every family member has a named job and one promise for the week.

Think about: Which job surprised someone by being fun?

The idea: a first version, tested by real users, then improved · Build your own

Do: Make the first version - a chore chart, a family recipe book, a plant-watering plan, a simple checklist made with the AI helper's coaching. Each person tries it for a few days. Change one thing based on what they found.

Copy this prompt:

"We made a first version of [our project]. Here's what we tried and what went wrong. Don't fix it for us. Ask us one question about each problem so we can fix it ourselves."

You'll know it worked when: Version two exists, and it changed because someone in the family used version one.

Think about: What did the youngest notice that the adults missed - or the other way round?

The idea: value - a builder asks who else it could help, and what makes something worth using · The builder economy

Do: Talk about who else might like what you built - a neighbor, a friend, a club. What would make it useful to them? Value is not only money: time saved, worry gone, a smile. Write one way it could help someone outside your home.

Copy this prompt:

"We built [project] for our family. Don't tell us how to sell it. Ask us two questions about who else it could help and what they would need."

You'll know it worked when: You named one person outside your home it could help, and what would make it useful to them.

Think about: What makes something worth someone's time?

The idea: measure, reflect, iterate - every good team looks back before going on · How evolution works

Do: Screens off. Everyone answers three questions round the table: what went well, what was hard, and what we'd do differently. Write the answers down and keep them for the next project.

Copy this prompt:

No AI this time - the family is the team that reflects. (Optional, afterwards with an adult: "Here are our three answers. Ask us one question about what we'd try next.")

You'll know it worked when: Everyone answered all three questions, and you kept the notes.

Think about: What did your family learn about working together?

The idea: master builders become mentors - teaching is the deepest learning · Global responsibility

Do: Invite a family you already know and trust (relatives, friends, neighbors) to try one Explorer lesson with you. Your children and elders lead parts of it. Let them do the guessing and checking themselves.

Copy this prompt:

"Our family will teach [a lesson] to friends who are new to AI. Don't write our plan. Ask us two questions about what they might find confusing or worrying, so we can prepare."

You'll know it worked when: Another family tried a lesson, led partly by your youngest and oldest.

Think about: What did you understand better after teaching it?

The idea: responsibility - our tools reach people we never meet, here and everywhere · Global responsibility

Do: Pick one AI tool your family uses. Together, list who it touches: you, the people who built it, people whose work it changes, people in other countries, and the planet (the electricity and devices it needs). For each, write one way it helps and one way it could cause a problem.

Copy this prompt:

"Here's our family's list of who this AI tool touches. Don't add to it. Ask us about one group we forgot and how they might see it."

You'll know it worked when: Your list has at least four groups, each with a help and a possible problem you thought of yourselves.

Think about: Is there one small thing your family could do differently because of this?

The idea: facing the future honestly and hopefully, from all sides · Why the world is changing

Do: Each person shares one hope and one worry about the world the youngest will grow up in. Pick one worry and test it: what makes it more likely, what makes it less likely? Then pick one hope and one small step the family can take toward it.

Copy this prompt:

"Our family is talking about this worry about the future: [worry]. Don't tell us not to worry, and don't agree just to please us. Ask us two questions about the evidence for it and two about the evidence against it."

You'll know it worked when: One worry tested from all sides, and one family step toward a hope, written down.

Think about: What do the oldest and the youngest agree on about the future?

The idea: planning for the long run - goals, small steps, measure, iterate · How evolution works

Do: Write a one-page family learning plan for the next year: one thing each person wants to learn, one project you'll build together, and one way you'll help someone else. Put a date on the calendar to look back and change the plan.

Copy this prompt:

"Here's our family learning plan for the year. Don't rewrite it. Ask us one question about each goal - is it small enough to start this month?"

You'll know it worked when: A one-page plan with something for everyone, and a date to review it.

Think about: What would you like your family to say about this year, a year from now?

The idea: AI as a patient explainer you can question · Build your own

Do: Ask a free AI helper to explain something you've always wondered about - a hobby, a recipe, how something works. Ask two follow-up questions.

Copy this prompt:

"Explain [something I'm curious about] simply. Then ask me one question to check I understood."

You'll know it worked when: You learned one thing you can explain to someone else.

Think about: How did it feel to have a patient explainer on call?

The idea: AI sounds confident even when it's wrong - checking is the skill · Build your own

Do: Ask it about something you know very well (your town, your job, your family's favorite recipe). Find one mistake or overconfident claim.

Copy this prompt:

"Tell me five facts about [something I know well]. Then tell me which one you're least sure of."

You'll know it worked when: You found or ruled out a mistake yourself.

Think about: What would you want your child to know about this?

The idea: AI that asks questions instead of giving answers · Build your own

Do: Sit with your child for one homework problem. Set the AI to tutor mode and let your child do the thinking.

Copy this prompt:

"Be a tutor for my child on [topic]. Don't give the answer. Ask one question at a time and give a hint only after two tries."

You'll know it worked when: Your child reached the answer and could explain it.

Think about: What did you notice about how your child thinks?

The idea: agreeing the rules together - what AI is for, and what we always do ourselves · How evolution works

Do: Write three family rules with your child: what AI may help with, what we always do ourselves, and what we never share.

Copy this prompt:

"Here are our three family rules for AI. Ask us one question about a situation they don't cover."

You'll know it worked when: Three rules on the fridge that your child helped write.

Think about: Which rule did your child suggest?

The idea: understanding the change together, without fear · Why the world is changing

Do: Read one story from our Daily or Your path forward page together. Talk about who it helps, who it leaves out, and what your child could build about it.

Copy this prompt:

"We read about [the change]. Don't tell us what to think. Ask us three questions to talk about as a family."

You'll know it worked when: You had one real conversation about the future - hopeful and honest.

Think about: What does your child hope for?

The idea: real feedback from a real person - you · The builder economy

Do: Use whatever your child built for a week. Tell them one thing that helped and one thing that confused you.

Copy this prompt:

"My child built [thing]. Help me phrase one encouraging and one honest piece of feedback. Don't write it for me - ask me what I noticed."

You'll know it worked when: Your child changed one thing because of your feedback.

Think about: What did you learn about your child as a builder?

The idea: agency - the project stays theirs · How evolution works

Do: For one week, answer every 'can you do it for me?' with a question back. Celebrate one thing they solved alone.

Copy this prompt:

"My child is stuck on [problem]. Give me three questions I can ask them instead of solving it."

You'll know it worked when: Your child solved one hard thing without you doing it.

Think about: How did it feel to hold back?

The idea: responsibility - the best builders ask who their work touches, near and far · Global responsibility

Do: With your child, pick their project or an idea they love. Together, list who it could affect - friends, other families, children in other countries, the planet. Let your child say how each might see it before you add anything.

Copy this prompt:

"My child and I made a list of who their project affects. Don't add to it. Ask us one question about someone we left out."

You'll know it worked when: Your child named at least one person or place they hadn't thought of before, by themselves.

Think about: What did your child say about people far away?

The idea: differentiation at speed - you choose and fix · Build your own

Do: Ask an AI to rewrite one reading passage or problem set at three difficulty levels. Fix what's wrong before you'd use any of it.

Copy this prompt:

"Rewrite this passage at three reading levels, keeping every fact. List anything you changed that I should check."

You'll know it worked when: One version you would actually use, after your fixes.

Think about: How much time did it save - and what did you have to correct?

The idea: see your assignment through a student's AI · Build your own

Do: Give one of your own assignments to an AI as a student would. See what it produces in one minute.

Copy this prompt:

"Complete this assignment as a strong student would." (Then ask:) "What parts of this would a teacher spot as not the student's own thinking?"

You'll know it worked when: You know which of your assignments an AI can do alone - and which it can't.

Think about: Which assignment would you redesign first?

The idea: grade the thinking, not just the answer · How evolution works

Do: Redesign one assignment so students show their process: a plan, a draft, what they changed and why, and one AI suggestion they rejected.

Copy this prompt:

"Here's my assignment. Suggest three ways to make students show their thinking process. Don't rewrite it; I'll choose."

You'll know it worked when: A redesigned assignment with at least two process checkpoints.

Think about: What will you learn about your students that you couldn't see before?

The idea: teach students to set AI's rules - struggle before answers · Build your own

Do: Teach one class period where students write their own tutor-mode prompt and use it on a hard problem. Collect one reflection from each.

Copy this prompt:

Students: "Be my tutor for [topic]. Don't give me the answer. Ask me one question at a time."

You'll know it worked when: Most students reached an answer by their own reasoning and can say how.

Think about: Which students surprised you?

The idea: talk about the future of work without fear or hype · Why the world is changing

Do: Use one story from our Daily or Your path forward page as a class discussion: who it helps, who it leaves out, what a young builder could do.

Copy this prompt:

"Give me five open discussion questions about [the change] for students aged [age]. Balanced, no answers."

You'll know it worked when: A discussion where students named both a worry and a possibility.

Think about: What do your students fear most - and hope for?

The idea: builders start from people · The builder economy

Do: Students interview someone in the school (a librarian, a coach, the office) about one weekly chore that wastes time, and pick one to solve.

Copy this prompt:

Students: "Here's what we heard. Ask us questions to understand the real problem. Don't suggest solutions yet."

You'll know it worked when: Each team has a one-sentence problem in the person's own words.

Think about: Which team found the most surprising need?

The idea: roles - one helper drafts, another checks, students decide · How evolution works

Do: Teams use one AI chat to draft and a second to review. Students decide every change, and log what they rejected and why.

Copy this prompt:

Reviewer: "List the three biggest problems in this work, most important first. Don't fix them."

You'll know it worked when: Each team's log shows at least one AI suggestion they rejected, with a reason.

Think about: Where did students show the best judgment?

The idea: teaching is how builders grow · Build your own

Do: Teams present what they built to the person they built it for, and teach another class how they did it.

Copy this prompt:

Students: "Help us plan a 3-minute demo. Ask us questions; don't write it for us."

You'll know it worked when: The person they built it for has used it at least once.

Think about: What would you change when you run it again?

The idea: global responsibility - students weigh their project's effects on others and the planet, from all sides · Global responsibility

Do: Each team maps who their project affects: the person it serves, people who don't use it, people in other countries, and the planet (devices, energy). Teams argue the strongest case for and against their project from at least two very different points of view before any AI is used.

Copy this prompt:

Students: "Here's our map of who our project affects. Don't fill it in. Ask us about one group we left out and how they would see it."

You'll know it worked when: Each team changed one thing in their project because of a view they hadn't considered.

Think about: Which team saw furthest beyond their own classroom?

The idea: long-horizon planning - teaching students to think twenty years ahead · Global responsibility

Do: Students write what their project, or the problem it solves, should look like in 2046, and one risk over twenty years. Each team commits to one action this term that makes the good version more likely.

Copy this prompt:

Students: "Here's our 2046 plan. Don't plan it for us. Ask us one question about a long-range risk and one about who should have a say."

You'll know it worked when: Every team has one long-range risk and one action for this term, written by the students.

Think about: What did your students hope for most?

The idea: AI as a study partner - it asks, you answer from your experience · Build your own

Do: Pick one thing everyone in your line of work ([your role]) must know (a code rule, a procedure, a safety step, a term). Write down your own explanation first, then let the helper quiz you on it.

Copy this prompt:

"My work: [your role]. Here's how I'd explain [one thing from my work]. Don't correct me yet. Ask me three questions a new trainee might ask, one at a time, then tell me which of my answers was weakest."

You'll know it worked when: You answered three questions from your own experience and found one thing to brush up on.

Think about: What do you know from experience that no manual says?

The idea: your experience is the check - AI sounds confident even when it's wrong · Build your own

Do: Ask the helper five questions about your work ([your role]) that you already know the answers to. Mark each answer right, wrong or half-right, and check one against an official source (a code book, a manual, a standard).

Copy this prompt:

"My work: [your role]. Tell me five facts about how [a common task] is done. Then tell me which one you're least sure of."

You'll know it worked when: You marked all five yourself and checked one in an official source.

Think about: Where would a beginner have been fooled?

The idea: privacy at work - describe the task, never paste the real names, numbers or secrets · Global responsibility

Do: Write two lists: what you must never put into any AI helper at work (customer details, patient information, passwords, prices, plans), and how to describe the same task in general words instead. Find your employer's AI rules, or ask your supervisor.

Copy this prompt:

"I want help with a task at work, described in general words: [general description]. Before you help, ask me one question to check I haven't included anything private."

You'll know it worked when: Two written lists, and you know what your workplace's AI rules are (or that there aren't any yet).

Think about: What would happen if something on your 'never' list leaked?

The idea: the gap - the same tool helps some tasks, threatens others, and changes what the job needs · Why the world is changing

Do: First, list the tasks in your week. Mark each: AI could help, AI could do it, AI should never do it. Then look up two real, dated examples of AI in your field - one that helps workers, one that worries them.

Copy this prompt:

"Here are my week's tasks, marked help / could do / never. Don't re-mark them. Ask me two questions about the evidence behind my marks, from all sides."

You'll know it worked when: Your marked task list, plus two real examples with sources you checked.

Think about: Which task do you most want help with - and which would you fight to keep?

The idea: start with busywork you do every week - reports, notes, quotes, schedules, emails · Build your own

Do: Pick one repeat paperwork task. Do it once yourself and time it. Then do it again with the helper (general words only), and time that. Compare the two for time and quality.

Copy this prompt:

"My work: [your role]. Here is a task I do every week, described in general words: [task]. Don't do it yet. Ask me three questions about what a good result looks like."

You'll know it worked when: Two timed versions, and an honest call on which was better and why.

Think about: What did the helper get wrong that you had to fix?

The idea: a spec - clear, reusable instructions make a helper consistent, like training a new colleague · Build your own

Do: Write one page of instructions for your task: the goal, the steps, the format, what to never do, and how you'll check it. Use it three times this week and improve one line each time.

Copy this prompt:

"Here are my instructions for a helper doing [task]. Don't rewrite them. Follow them once on this made-up example, then tell me which instruction was unclear."

You'll know it worked when: A one-page instruction sheet you've used three times and improved.

Think about: What did writing it down teach you about your own job?

The idea: verification - an AI-drafted checklist is only trusted after a person and an official source agree · Build your own

Do: Draft a quality or safety checklist for one job in your work ([your role]) from your own experience. Ask the helper what you might have missed, then check every item against the official source or with your supervisor before you use it.

Copy this prompt:

"Here is my checklist for [a job]. Don't add items. Ask me about two situations my list might not cover, and I'll decide."

You'll know it worked when: A checklist where every item is verified by you and an official source or supervisor.

Think about: Which item did only your experience catch?

The idea: measure, then decide - keep what helps, drop what doesn't · How evolution works

Do: For two weeks, write one line each time you use your helper: minutes saved (or lost) and any mistake you caught. At the end, decide: keep it, change it or drop it.

Copy this prompt:

"Here is my two-week log. Don't decide for me. Ask me two questions that help me judge whether it's really worth it."

You'll know it worked when: A two-week log and a written keep / change / drop decision.

Think about: What surprised you in the numbers?

The idea: orchestration - a shared helper with shared rules, agreed with the people who use it · Build your own

Do: With your supervisor's OK, share your best instruction sheet with two coworkers. Agree three team rules (what it's for, what never goes in, who checks the result). Collect their changes after a week.

Copy this prompt:

"Here are my team's three rules and our instruction sheet. Don't change them. Ask us about one situation the rules don't cover yet."

You'll know it worked when: Two coworkers used it for a week, and the sheet changed because of them.

Think about: What did your coworkers see that you didn't?

The idea: master builders become mentors - teaching is how a whole team rides the wave · The builder economy

Do: Sit with one coworker who hasn't tried AI and walk them through the first two lessons of this path, their way, on their own task. Let them do the typing and the deciding.

Copy this prompt:

"I'm teaching a coworker who is new to AI. Don't write my plan. Ask me two questions about what might worry or confuse them."

You'll know it worked when: Your coworker finished one lesson themselves.

Think about: What did you understand better after teaching it?

The idea: global responsibility - workplace AI touches coworkers, customers, apprentices, the public and the planet · Global responsibility

Do: List everyone your team's AI use touches: coworkers, apprentices who need to learn the basics, customers or patients, the public who relies on your work being safe, and the planet (the electricity it needs). Write one way to make it fairer or safer for each.

Copy this prompt:

"Here is my list of who our AI use touches. Don't add to it. Ask me about one group I forgot and how they might see it."

You'll know it worked when: A list of at least four groups, each with one change you could make.

Think about: How do apprentices still learn the basics if AI does the easy parts?

The idea: careers - roles evolve; the people who map the change get to choose their next rung · The builder economy

Do: Write how your work ([your role]) might look in five years: which tasks shrink, which grow, what new skills it needs. Mark one skill to start learning this month, and one person who could teach you. (If your job is changing fast, the worker path's next-rung lessons are made for you.)

Copy this prompt:

"Here's my picture of my role in five years. Don't predict it for me. Ask me two questions about the evidence from all sides, and one about the first step I could take this month."

You'll know it worked when: A one-page five-year picture with one skill and one first step chosen.

Think about: What part of your work do you want to grow into?

The idea: your experience - people, place, craft, judgment - is the asset you carry into any new work · Why the world is changing

Do: On paper, before any AI, list ten things you know from your work: problems you can spot, people you know how to help, tricks of the trade, safety habits. Circle the three that took you longest to learn.

Copy this prompt:

"Here is my list of what I know from my work. Don't add to it. Ask me three questions, one at a time, that help me find skills I left out."

You'll know it worked when: Your list grew by at least two items that you found yourself.

Think about: Which item on your list would be hardest for a machine to learn?

The idea: AI as a helper on a task - the judgment stays yours · Build your own

Do: Pick one task from your work or daily life that eats your time (a schedule, a checklist, a letter). Write your own first attempt, then ask a free AI helper to improve one part. Keep only what you agree with. Use made-up names and numbers, never real customers or coworkers.

Copy this prompt:

"Here is my first draft of [task]. Don't rewrite it. Point out the one weakest part and ask me how I'd fix it."

You'll know it worked when: You fixed one part yourself after the helper's question, and you can say why you kept or rejected its advice.

Think about: Where did your own experience beat the helper's suggestion?

The idea: AI sounds sure even when it's wrong - your experience is the check · Build your own

Do: Ask a free AI helper five questions about your own trade or job that you already know the answers to. Mark each answer right, wrong, or half right.

Copy this prompt:

"I'm going to test you on [my trade]. Answer my five questions briefly, then tell me which answer you're least sure of."

You'll know it worked when: You found at least one mistake or overconfident answer yourself - or ruled them all out.

Think about: What would someone without your experience have believed?

The idea: the gap analysis - the change hits tasks, not whole people, and new tasks grow where the bottlenecks are · Why the world is changing

Do: Write the main tasks of your job (or your last job) in general words - 'scheduling', 'inspections', 'answering customer questions' - never your employer's name or anything confidential - in two columns: tasks machines are starting to do, and tasks that still need a person. Then read one story about your industry from our Daily or a news source and move anything that belongs in the other column.

Copy this prompt:

"Here are my two columns of general job tasks (no company names or private details). Don't move anything. Ask me two questions that test whether a task is really in the right column."

You'll know it worked when: Both columns are filled, and at least one task moved because of your own reasoning.

Think about: Which growing task is closest to what you already do well?

The idea: a change of work is a project - budget the hours and energy you really have, and protect rest and the people you care about · The builder economy

Do: On paper, write your real week: work or job search, family, rest, the things that keep you well. Find the hours you can honestly give to learning - even two is a start. Make one promise to yourself you can keep, and one thing you won't give up for it. Keep the plan on paper; share only rough hours per category with the helper - no names, places or appointments.

Copy this prompt:

"Here are my rough hours per week for work, family, rest and learning - no details. Don't tell me to do more. Ask me two questions about whether it's realistic and one about what keeps me well."

You'll know it worked when: A week plan with learning time you kept, and one thing you protected.

Think about: What does a sustainable pace look like for you right now?

The idea: career paths at any age - more school or training, the trades and technical work of the build-out, or building your own; each is real, and many people combine them · Career paths

Do: On paper, draw three roads from where you are now: more school or a training program, the trades and technical work that build power, factories and data centers, and building something of your own. Under each, write what you already bring from your work, what you'd need to learn, and one thing that worries you. Look up one real job on each road.

Copy this prompt:

"Here's my map of three roads from where I am now. Don't pick one for me. Ask me two questions about each road, from all sides - what it would ask of me, and what I already bring."

You'll know it worked when: A three-road map with one real job on each road and one skill you already have that fits it.

Think about: Which road uses the most of what you already know?

The idea: the closest new niche reuses the most of what you already know · The builder economy

Do: From these niches - maintaining automated lines and systems; the build-out trades (electrician, line worker, HVAC, construction, technician); teaching and supporting others through the change; a small local service AI makes affordable to start - pick the two closest to your list from lesson w1-1. Write three reasons for each, then choose one.

Copy this prompt:

"I'm choosing between [niche A] and [niche B]. Here are my reasons. Don't choose for me. Ask me the three questions a career counselor would ask before I decide."

You'll know it worked when: You chose one niche and can say in one sentence why it fits you.

Think about: What part of your old work comes with you into the new niche?

The idea: learning a new skill with a patient tutor - you do the thinking, it asks the questions · Build your own

Do: Pick the first skill your niche needs (reading a wiring diagram, basic electrical theory, a maintenance log, a spreadsheet, explaining a tool to a beginner). Spend 40 minutes learning it with the AI in quiz mode, writing your answers before it responds.

Copy this prompt:

"Teach me the basics of [skill] for a beginner. Don't lecture. Ask me one question at a time, wait for my answer, and give a hint only after I've tried twice."

You'll know it worked when: You can explain the basics of the skill to someone else without notes.

Think about: What was easier than you expected?

The idea: most trade and technician paths have a free or paid-while-you-learn way in - apprenticeships, community colleges, public workforce centers · The builder economy

Do: Look up, yourself, the free or paid-while-you-learn ways into your niche near you: apprenticeships, community college programs, your public workforce or job center. Write down three, with what each asks of you to start.

Copy this prompt:

"I'm looking into [niche] near where I live. Don't search for me. Help me write five questions to ask an apprenticeship office or job center so I learn what the first step really costs and needs."

You'll know it worked when: You have three real options written down, and five questions ready for one of them.

Think about: Which option fits your life right now - and what would you need to try it?

The idea: evolution in practice - a goal, a measure, try, check, adjust · How evolution works

Do: Set one small goal for the next two weeks (an hour of practice a day, one call to a program, one conversation with someone in the niche). Pick one number to track. At the end, write what you'll change.

Copy this prompt:

"Here is my two-week goal and my number. Ask me what could get in the way, and what I'd do then. Don't make the plan for me."

You'll know it worked when: Two weeks of your number written down, and one change for the next two weeks.

Think about: What did the number tell you that your feelings didn't?

The idea: you tell your own story; the helper is your critic, not your ghostwriter · Build your own

Do: Write a short skills summary for your new niche in your own words: what you've done, what you know, what you're learning. Leave out your address and contact details. Then ask the helper to critique it.

Copy this prompt:

"Here is my skills summary for [niche]. Don't rewrite it. Tell me the two places where a hiring manager would want more, and ask me for the real example."

You'll know it worked when: Your summary has at least two concrete examples from your own work, written by you.

Think about: Which example are you proudest of?

The idea: budgeting a career move - start with free tools and free or paid-while-you-learn paths; plan any later cost before you ever spend · The builder economy

Do: List what your niche asks for: skills, tools, certificates. For each, find a free way to start (library, public programs, free courses, apprenticeships that pay you). If something truly costs money later, write it as a plan: what it would return, and whether a program or employer usually covers it. You don't buy anything in this lesson.

Copy this prompt:

"Here's my list of what the niche needs and my free ways in. Don't recommend anything to buy. Ask me two questions that help me find a free option I missed, and one about whether a later cost would really pay back."

You'll know it worked when: Every item on your list has a free way to start, and any later cost is written as a plan with who usually pays it.

Think about: Which free way in surprised you most?

The idea: a small agentic helper with one clear job - spec first, you check its work · Build your own

Do: Write a one-page spec for a helper your crew or group could use (a shift checklist, a maintenance log summary, a safety reminder list): its goal, what it must never do, and how you'll test it. Build a first version in a free AI chat using made-up sample data only.

Copy this prompt:

"Here is my spec for a crew helper. Don't build it yet. Find the gaps: what's unclear, what could go wrong, and what test I'm missing."

You'll know it worked when: Your helper passes the tests you wrote, on made-up data, and you fixed at least one gap yourself.

Think about: What decision did you keep for a person, not the helper?

The idea: every person who helps another across makes the change smaller for both · The builder economy

Do: Walk one coworker or friend through lesson w1-1 (what you know that a machine doesn't). Let them write their own list - you ask the questions.

Copy this prompt:

"I'm helping a friend map their skills. Give me five good questions to ask them. Don't suggest answers."

You'll know it worked when: Your friend finished their own list and picked one next step.

Think about: What did you see in them that they didn't see in themselves?

The idea: AI makes some small local services newly affordable to start - at $0, with what you have · The builder economy

Do: Sketch one small service you could offer with your experience and free tools (repair advice, setting up neighbors' devices, a local trade guide). Write who it helps, what you'd do first for free, and how you'd know it's useful.

Copy this prompt:

"Here is my idea for a small service. Don't improve it. Ask me the three hardest questions a first customer would ask."

You'll know it worked when: A one-page sketch with a free first step you could take this week.

Think about: Who in your community needs this most?

The idea: global responsibility - one change at work reaches coworkers, families, towns, other countries and the planet · Global responsibility

Do: Pick one change at your work (a new machine, a new AI system, a new data center or factory nearby). On paper, map who it affects: your crew, customers, families, your town, workers in other countries, and the land, water and power it uses. For each, write one way it helps and one way it could hurt.

Copy this prompt:

"Here is my map of who this change affects. Don't fill it in. Ask me about one group or place I left out, and what they would say from all sides."

You'll know it worked when: Your map has at least five groups, each with a help and a harm you wrote yourself.

Think about: What is one small thing you could do to make the change fairer for the group it hurts most?

The idea: long-horizon planning - thinking past this year's job to the work your town will need in ten years · Global responsibility

Do: Write what work in your town might look like in ten years: what will be built, what will need maintaining, what people will need taught. Then write three things you, your crew or your town could start now so people are ready - including the young people coming up.

Copy this prompt:

"Here is my ten-year picture of work in my town. Don't predict for me. Ask me two questions that test my assumptions from all sides, and one about who my plan might leave out."

You'll know it worked when: A one-page ten-year plan with three actions that could start this year.

Think about: What would you tell a young person in your town who is choosing their first job today?

The idea: honesty and credit - be open about how you use AI, credit the people who taught you, and look after your crew · The builder economy

Do: Write a short note on how you use AI helpers in your work: what they do, what you always check, and who taught you. Then write one thing you'll do this month to look after your crew through the change - teaching, listening, or speaking up for someone.

Copy this prompt:

"Here's my note on how I use AI and how I'll look after my crew. Don't rewrite it. Ask me whether anyone deserves credit I left out, and whether my promise is one I can keep."

You'll know it worked when: Your note is honest about the AI's part, credits at least one person, and has one promise you keep this month.

Think about: Who taught you the most at work, and how could you pass it on?

The idea: use the helper to build understanding, not to skip it

Do: Pick a small piece of code you didn't write (a function in an open-source project or a tutorial). Explain what it does, line by line, in your own words, before asking the AI anything.

Copy this prompt:

"Here's my explanation of this function. Don't explain it to me. Ask me questions about the parts I got wrong or skipped, one at a time."

You'll know it worked when: Your explanation survives three of the helper's questions without changes - or you fixed it yourself.

Think about: Which line did you think you understood but didn't?

The idea: directing a helper - you set how much it gives you

Do: Solve a small coding exercise (a kata or a homework problem) with the helper in hint-only mode.

Copy this prompt:

"I'm solving [problem] in [language]. Don't write code for me. When I'm stuck, give me the smallest hint that gets me moving, and ask what I've tried."

You'll know it worked when: Your solution passes its tests, and you wrote every line.

Think about: What would you have missed if the helper had just written it?

The idea: debugging is reasoning - the helper checks your reasoning, not replaces it

Do: Break a working program on purpose. Read the error, write your guess at the cause, then ask the helper to check your reasoning.

Copy this prompt:

"Here's the error and my guess at the cause. Tell me only whether my reasoning holds, and what evidence would prove it."

You'll know it worked when: You found the cause yourself and can say how you proved it.

Think about: What did the error message tell you that you first ignored?

The idea: reviewing AI code - plausible is not correct

Do: Ask the helper for a function with a tricky edge case (dates, empty lists, unicode). Write tests that try to break it.

Copy this prompt:

"Write a function that [task]. Then list the edge cases you're LEAST sure you handled."

You'll know it worked when: You wrote at least three tests and found or ruled out a bug yourself.

Think about: Did its own list of weak spots match what your tests found?

The idea: the spec - goal, constraints, and how you'll know it worked - is the most valuable thing you write

Do: Before any code, write a one-page spec for a small tool someone you know needs: the goal, who it's for (a role like 'a club organizer', never a name or personal details), what it must not do, and the acceptance tests in plain words. Use made-up sample data, never real people's data.

Copy this prompt:

"Here's my spec. Don't write code. Find ambiguities, missing edge cases and tests I can't actually check."

You'll know it worked when: Someone else can read the spec and predict what the tool will do.

Think about: Which question from the review changed your design?

The idea: tests are the contract an agent must meet

Do: Turn your acceptance tests into automated tests. They should all fail now.

Copy this prompt:

"Here are my plain-language tests. Help me turn ONE into an automated test, explaining each part, then I'll do the rest."

You'll know it worked when: You have a failing test suite that matches your spec.

Think about: Which test was hardest to make precise?

The idea: delegate a well-scoped task, then review every line

Do: Give a coding agent one small task from your spec in a throwaway branch. Review its change like a colleague's: read every line, run the tests, ask for fixes.

Copy this prompt:

"Implement [one task] to make these tests pass. Keep the change small. Explain every decision I should check."

You'll know it worked when: The tests pass and you can explain every line you merged.

Think about: What did you reject, and why?

The idea: feedback loops - real use beats your own opinion

Do: Put version 1 in front of the person it's for. Watch, don't help. Write down where they got stuck and fix the biggest one.

Copy this prompt:

"In general terms - no names, no personal details, no real user data - here's where my user struggled. Ask me which fix matters most to them - don't pick it for me."

You'll know it worked when: Version 2 fixes the biggest problem your user hit.

Think about: What did they need that your spec never said?

The idea: separate roles catch what one agent misses

Do: Run one feature with three separate agent sessions: a builder, a reviewer that only critiques, and a tester that only writes tests. You arbitrate.

Copy this prompt:

Reviewer: "You are a strict reviewer. List the three most serious problems in this change, most important first. Do not fix them."

You'll know it worked when: The reviewer or tester caught at least one real problem the builder missed, and you decided what to do about it.

Think about: Which disagreement between agents did you settle, and how?

The idea: every agent needs an explicit goal and a measurable bar - iterate or hold on a miss

Do: For your project, write one goal and two measures with thresholds (e.g. tests passing, time to fix a bug, a user task completed). Gate your merges on them.

Copy this prompt:

"Here are my goal and thresholds. Tell me how an agent could pass them while missing the real goal."

You'll know it worked when: Your merges are gated on measures, and one merge was held because it missed.

Think about: What would make your measure look good while the goal isn't met?

The idea: plan, delegate, measure, improve - every week

Do: Run a one-week sprint with your agents: plan on Monday, delegate scoped tasks, measure on Friday, write one process improvement.

Copy this prompt:

"Here's my sprint: plan, results, measures. Ask me two questions that help me pick ONE process change."

You'll know it worked when: One full loop done, with numbers and one written improvement.

Think about: Where did you spend your own time - and where should you have?

The idea: judgment stays human: safety, privacy, ethics, credit, what to build at all

Do: Write your team's rules: what agents may do alone, what needs your review, and what they never do (secrets, production data, irreversible actions).

Copy this prompt:

"Here are my rules for my agent team. Ask me about one risky situation they don't cover."

You'll know it worked when: Your rules cover secrets, irreversible actions and anything touching people's data.

Think about: Which rule would you most regret not having?

The idea: every builder who teaches makes two more builders

Do: Teach a newer developer the spec-first, builder-reviewer-tester method on their own small project.

Copy this prompt:

"Help me plan a 30-minute session teaching spec-first agent work. Ask me what they already know first."

You'll know it worked when: They ran one builder-and-reviewer round on their own.

Think about: What did teaching it show you about your own method?

The idea: global responsibility - before you ship, ask who it reaches and what it costs the world · Global responsibility

Do: Before your next release, write a one-page impact review yourself: who uses it and who it affects without using it, how it could be misused, who it excludes (accessibility, language, cost, slow connections), and what compute and energy it uses. Then have a reviewer agent challenge it.

Copy this prompt:

Reviewer: "Here is my impact review. Don't rewrite it. Name the two groups or harms I'm most likely to have missed, from all sides, and ask me how I'd measure them."

You'll know it worked when: Your review names at least five affected groups, and you changed one thing in the release because of it.

Think about: Which effect of your software reaches people you will never meet?

The idea: long-horizon planning - software and agent systems outlive their builders; plan who carries them and what they cost over years · Global responsibility

Do: For a system you run, write a ten-year view: what will need maintaining, what breaks if you step away, what data it keeps and for how long, and how its energy and compute use grows. Write one change you make this month so it can be handed on safely.

Copy this prompt:

"Here's my ten-year view of my system. Don't plan it for me. Ask me the two questions a person inheriting it in five years would most want answered."

You'll know it worked when: A written hand-on plan, and one change merged this month that makes the system easier to hand on or cheaper to run.

Think about: Who will carry what you build after you - and what do you owe them?

The idea: AI as a patient explainer you can question as much as you like · Build your own

Do: Before you ask, write down what you already think the answer is. Then ask a free AI helper one question you've always wondered about - how something works, a bit of history, a garden problem - and ask two follow-up questions.

Copy this prompt:

"Here is what I think the answer is. Don't just tell me the answer. Tell me which part of my guess is right, then ask me one question to help me work out the rest."

You'll know it worked when: You can explain the answer to a friend in your own words.

Think about: How close was your first guess?

The idea: AI sounds sure even when it's wrong - your knowledge is the check · Build your own

Do: Ask it about something you know very well - your town's history, your old line of work, a hobby. Find one mistake or overconfident claim yourself.

Copy this prompt:

"Tell me five facts about [something I know well]. Then tell me which one you're least sure of."

You'll know it worked when: You found a mistake yourself - or checked them all and found none.

Think about: How would you check it on a subject you don't know well?

The idea: the same tools that help can be used to fool people - knowing the signs is your protection · Build your own

Do: Write down every sign of a scam call or message you can think of first. Then compare your list with the helper's questions, and agree a family code word or a call-back rule with someone you trust.

Copy this prompt:

"Here is my list of signs that a call or message is a scam. Don't give me a list. Quiz me with three made-up messages, one at a time, and let me say whether each is a scam and why."

You'll know it worked when: You spotted the scams in the quiz and have one call-back rule agreed with someone you trust.

Think about: What would you tell a friend to do if a familiar voice asked for money?

The idea: taking a worry seriously means testing it - looking at the evidence from all sides · Why the world is changing

Do: Write down one worry you have about AI, in your own words. Write what would make it more likely and what would make it less likely. Then ask the helper to question you.

Copy this prompt:

"Here is my worry about AI and my reasons. Don't tell me not to worry and don't agree just to please me. Ask me two questions about the evidence for it and two about the evidence against it."

You'll know it worked when: You wrote one thing that makes your worry more likely and one that makes it less likely - both checked against a trusted source.

Think about: Is your worry bigger, smaller or different now? (If you like, tell our panel about it on our Tell us page - no personal details needed.)

The idea: the helper is your editor, never your ghostwriter · Build your own

Do: Write a letter you've been meaning to write (a thank-you, a note to a club, a letter to your local paper). Write the whole first draft yourself. Use [name] in place of real names and leave out addresses.

Copy this prompt:

"Here is my letter. Don't rewrite it. Tell me the one sentence that's least clear and ask me what I meant."

You'll know it worked when: You improved your letter yourself, and it still sounds like you.

Think about: What did you change, and why?

The idea: a patient coach for something you love - you practice, it asks · Build your own

Do: Pick a hobby (gardening, a language, woodworking plans, music, family recipes). Choose one next step you'd like to learn and try it, with the helper coaching instead of telling.

Copy this prompt:

"I want to get better at [hobby]. Ask me what I already do, then give me one small challenge to try this week. Give hints only if I ask."

You'll know it worked when: You tried the challenge and can say what you learned from it.

Think about: What would you like to learn next?

The idea: every person who helps another get started makes the change kinder for both · The builder economy

Do: Sit with a friend or neighbor who's never tried an AI helper. Let them type; you only ask questions. Share your scam rules from lesson c1-3.

Copy this prompt:

"I'm helping a friend try AI for the first time. Give me three good first questions they could ask about something they love. Don't pick the topic for them."

You'll know it worked when: Your friend asked their own question and checked one answer.

Think about: What did your friend find surprising?

The idea: evolution in practice - try, measure, adjust · How evolution works

Do: Pick one small thing to try with AI each day for a week (learning a few words of a language, planning meals, one question a day). Keep a simple tally. At the end, decide: keep, change, or stop.

Copy this prompt:

"Here's my one-week habit and my tally. Ask me whether it was worth it and what I'd change. Don't decide for me."

You'll know it worked when: A week's tally and a decision you made yourself.

Think about: Was it worth your time? Why?

The idea: mentoring and local history - the helper interviews, you tell the story · The builder economy

Do: Choose one thing you know that should be passed on (how your town has changed, a skill from your working life, lessons learned the hard way). Let the helper interview you, then write it up in your own words. Leave out other people's private details.

Copy this prompt:

"Interview me about [what I know]. Ask one question at a time and follow up on what I say. Don't write my story for me."

You'll know it worked when: One page in your own words that a younger person could learn from.

Think about: Who would you most like to read it?

The idea: a small helper with one clear job - spec first, you check it · Build your own

Do: For a group you belong to (a club, a congregation, a garden group), write down one job a helper could do: answering common questions, a sign-up reminder list. Write what it must never do and three test questions. Try it in a free AI chat with made-up details only.

Copy this prompt:

"Here is the job for my group's helper and my three tests. Don't build it yet. Tell me what's unclear and ask me how it should answer a question it doesn't know."

You'll know it worked when: Your helper answers your three tests well, and says 'I don't know' when it should.

Think about: What should a person in your group always answer, not a helper?

The idea: global responsibility - a change near you reaches people far away and the planet too · Global responsibility

Do: Pick one AI change you've read about (a new data center, AI in hospitals, AI in schools). On paper, write who it affects - near you and far away, now and in twenty years - and the land, water and power it uses. For each, one way it helps and one way it could hurt.

Copy this prompt:

"Here is my list of who this change affects. Don't fill it in. Ask me about one group or place I left out, and how they would see it from all sides."

You'll know it worked when: At least five groups or places, each with a help and a harm you wrote yourself.

Think about: What would make this change fairer for the people it could hurt most?

The idea: long-horizon planning - what we do now shapes the world the next generation inherits · Global responsibility

Do: Write a letter to the young people of 2046. Say what you hope their world is like, what you worry about, and three things people today should start doing so the hopes win. Use no real names.

Copy this prompt:

"Here is my letter to 2046. Don't rewrite it. Ask me two questions that test my hopes and worries from all sides, and one about what I could start doing myself this year."

You'll know it worked when: A one-page letter with three things to start now, and one you will do yourself.

Think about: What did you learn in your own life that you most want them to know?