Build the machine that builds the machine

You've done this before

Last week you asked for help with a math problem, and this week you asked again. It was the same kind of problem, just with different numbers.

Maybe it isn't math for you. Maybe it's spelling words every Friday, or asking an AI for a new story idea every single day.

So here's a question. How many times have you done the exact same task twice? Each time, you started from zero. What if the second time could be almost instant?

The big idea

Think about a video game that comes with a level editor. Playing one level is fun, but a player with the level editor can create a hundred levels.

Or think about baking cookies. You could shape every single cookie by hand. Or you could spend five minutes making a cookie cutter, and after that each cookie takes about one second.

Builders have a saying for this idea: build the machine that builds the machine. Don't just make the thing you need. Make the tool that makes the thing, and then use that tool as many times as you want.

There is a catch, though. Building the tool is slower the first time, because shaping one cookie by hand is quicker than making a cutter. But by the tenth cookie, the cutter is winning, and by the hundredth cookie it isn't even close.

So here's the trade: a machine is slower at first and much faster forever after.

Chat mode vs. machine mode

Lots of people use AI like a question box. You type a question, get an answer, and close the window. Next time, you start all over again. Let's call that chat mode.

There's another way to use it. You ask the AI to build you a tool, and then you use that tool again and again. Let's call that machine mode.

A prompt is the message you type to the AI. An engine is a tool that does the same kind of job every time you run it. Your engine might be a saved prompt, a spreadsheet, a printable, or a small program. What matters is that you can run it again tomorrow.

Chat mode	Machine mode
Ask a question	Ask AI to build a tool
Get one answer	Test the tool
Chat closes, the answer is gone	Save the tool
Next time, start over	Next time, run the tool again

In machine mode, the AI helps you build something you get to keep. You own it, you can improve it, and you can even share it with a friend.

For parents and teachers: leverage

An answer gets used once. A tool gets used again and again, and each run costs almost nothing. That's leverage: effort spent once that keeps paying back. It's why adults write recipes, checklists, and templates instead of solving the same problem fresh every time. For a child, the bigger payoff is the habit behind it. A kid who asks "will I need this again?" starts to see their work as something they can improve, not just finish. That question is useful far beyond AI.

A worked example

Imagine you're stuck on two-digit multiplication, like 34 × 27. Here are two different ways to ask an AI for help.

Weak prompt (chat mode)	Strong prompt (machine mode)
"Teach me two-digit by two-digit multiplication."	"Build a math practice engine that understands K–12 math and can make practice problems for any topic I choose. Prove it works by using it to make a worksheet that teaches two-digit by two-digit multiplication."

The weak prompt isn't bad, and you'll get a decent explanation. But look at what happens next.

Tomorrow: You still need more practice. With the weak prompt, you have to ask all over again. With the strong prompt, you just ask your engine for ten more problems, and they can be harder ones if you want.

Next week: Your class moves on to fractions. The weak prompt can't help, because it only knew about multiplication. Your engine can make fraction problems in a few seconds.

Next year: Algebra shows up. The weak prompt is long gone. Your engine is still there, ready for whatever topic you choose.

Notice the end of the strong prompt: "Prove it works by…" That line is important. It makes the AI show you real results right away, so you can see whether the engine actually works. You also get the worksheet you needed today.

The recipe

Here's a template you can reuse. A template is a fill-in-the-blank pattern for writing something.

The Machine Recipe

Build me a [tool] that can [do a general job that covers lots of cases]. Prove it works by using it to [do one specific thing I need right now].

The middle blank is the most important part. Make it big enough that the tool will still be useful next week.

Here it is in three other places:

Vocabulary: "Build me a quiz maker that can turn any word list into practice quizzes. Prove it works by using it to quiz me on this week's 12 spelling words."
Sports: "Build me a practice planner that can make drills for any soccer skill and any amount of time. Prove it works by using it to plan 20 minutes of dribbling drills."
Stories: "Build me a story-starter machine that can mix any setting, character, and problem. Prove it works by using it to give me three ideas for a spooky story."

For parents and teachers: sizing the middle blank

The hard part of the recipe is the middle blank. To fill it, a child has to look at one problem and name the pattern behind it: not "these 12 words" but "any word list." That move from example to category is the heart of systems thinking. It's also where kids overreach. "An engine for all of school" is too vague to test or trust. Help them size the blank: big enough to reuse next month, small enough that they could tell whether it's working.

When NOT to build a machine

Machines aren't always the right choice. Here are two rules.

Rule 1: If you'll only do it once or twice, just do it yourself. Making a cookie cutter for one cookie is a waste of time. If you need one birthday card, just write the card. Build a machine when you can see the same job coming back again.

Rule 2: If the point is for your brain to learn something, the machine should give you practice, not do the thinking for you. A math engine that makes problems for you to solve is great, because it gives your brain more practice. A machine that does your homework is the opposite, because it skips the part that was supposed to make you stronger.

Here's a quick way to check. Ask yourself, "After this, could I do it without the machine?" If the answer is yes, or closer to yes than before, you built the right kind of machine. If the answer is no, you built a shortcut around your own brain.

For parents and teachers: practice machines vs. thinking machines

The line between productive AI use and outsourced thinking isn't whether AI was involved. It's where the effortful thinking happened. A practice-problem generator raises the number of times a child recalls and applies a skill, which is the part that makes learning stick. A machine that hands over finished answers removes that step. One question sorts almost every case: "What did the machine do, and what did you do?" If the child's half is "I solved them," the tool is working. If it's "I copied them," it isn't. Ask it often and without judgment, and kids start asking it themselves.

Check your machine

AI can be wrong, and sometimes it's wrong while sounding completely sure of itself. That means a machine built with AI can have bugs. A bug is a mistake hidden inside a tool.

Your math engine might print an answer key with a wrong answer. Your quiz maker might misspell one of the words. Your practice planner might plan 40 minutes of drills when you asked for 20.

That's why real builders test their machines. Here's how:

Check a few answers by hand. Work three answers out yourself and compare.
Try tricky inputs. Ask for something unusual, like zero, a really long word, or a topic the machine hasn't seen yet.
Fix it when it breaks. Tell the AI exactly what went wrong, and ask it to fix the engine instead of just that one answer.

Testing isn't extra work. It's part of building. Anyone can ask for a machine, but a builder is the person who makes sure it works.

For parents and teachers: verification is the skill

AI produces fluent, confident output whether or not it's correct. That makes checking work as basic a skill as producing it. Building a machine gives kids a natural reason to check: it's their tool, and a broken tool reflects on its builder. Encourage spot-checks by hand, strange inputs, and the question "how would I know if this were wrong?" One practical note: many AI tools require users to be 13 or older, and some need a parent's consent beyond that. Check each tool's terms. With younger kids, build the machine together or supervise its use.

Your challenge for today

Think of something you've asked an AI more than once, or something you do by hand every single week. It could be spelling practice, a warm-up routine, a chore chart, or story ideas.

Today, build the machine instead. Use the recipe:

"Build me a ___ that can ___. Prove it works by using it to ___."

Then test your machine before you trust it. Check three results by hand, try one unusual input, and fix whatever breaks. Save it where you'll find it next week. (If you're under 13, do this with a parent or teacher.)

Next week, run your machine again. That's the moment it clicks, because the work you did once just paid you back.

Askers start over every time. Builders pick up right where they left off.