I Asked Claude to Build Worksheets for My Kids. Now I Can't Go Back.

My daughters are seven. They are at the stage where multiplication tables need to become automatic, the kind of recall that just fires without effort. So I did what most parents do. I opened a browser tab and started looking.

Forty minutes later, I had three browser tabs open, two PDF previews that required me to disable my ad blocker, one worksheet pack that was clearly designed for a different curriculum, and nothing I actually wanted to print. This is the dirty truth about sourcing educational materials as a parent. It is not that good resources don’t exist. It is that finding them is genuinely tedious, and what you eventually settle for is rarely what you actually needed.

The problem isn’t just quality. It’s the time it takes to find out the quality is wrong.

The worksheet packs you buy online are built for an average child at an average stage. That is not a criticism of the people who make them. It is just a structural limitation. A pack designed for “Year 2 multiplication” has to work for every Year 2 child, which means it will be too easy for some, too hard for others, and slightly misaligned for almost everyone.

You don’t find this out when you buy it. You find it out when your kid sits down and either breezes through it without thinking or stalls at question four. By then you’ve already spent the money, and more importantly, you’ve already spent the time.

The internet version of this problem is worse. Free worksheets come with ads, with forced signups, with previews that don’t match what downloads, and with the lingering feeling that you’re one wrong click away from installing something on your laptop. I’ve spent longer trying to find a usable free worksheet than it would have taken me to just type the questions out myself.

When customisation becomes cheap, the whole calculation changes

I work in enterprise AI enablement. My job is helping organisations figure out where AI actually creates value and where it is just noise. The answer almost always comes back to the same principle: AI’s biggest unlock is not speed. It is making customisation economical.

This clicked for me in a personal way when I tried asking Claude to build multiplication worksheets. I gave it specific parameters. A4 format, three columns, 20 rows per column, so 60 questions per page, randomised from 2 times 2 through to 9 times 12, with a name field and score box at the top. It took about 30 seconds to generate a preview.

Then I pushed further. My daughters are twins, and if they get identical sheets, they compare answers instead of thinking. So I asked for two separate sets, different names at the top, completely randomised independently. I also asked for about 5% of questions to involve 0s and 1s, the fundamentals that standard packs often skip because they feel too easy, but that kids still need to have solid.

Thirty sheets per child. Enough for a month of daily practice. The whole process took maybe ten minutes.

For parents of one child, the personalisation still applies. You can tell Claude what your child has already covered and ask it to avoid repeating those question patterns. You can weight certain tables more heavily if you know those are the gaps. You can adjust difficulty as the weeks go on. The worksheet evolves with the child, not the other way around.

Maths is just where I started

There is a homework assignment my daughters get before school holidays that I suspect every parent with school-age children has encountered. Write and perform a short skit or presentation. The intention is good. It builds confidence, it’s creative, it’s memorable. The reality is that seven-year-olds cannot write a skit with any meaningful structure on their own. So it becomes parent homework, which is fine, but it used to take me a disproportionate amount of time.

Now I describe the theme, the characters, my daughters’ personalities, the rough length the teacher expects, and the kind of humour that lands with kids their age. Claude drafts a script. I read it, adjust a few lines to match how my daughters actually speak, and we rehearse from there. What used to take an evening now takes about 20 minutes.

I have not yet tried this across every subject, and I am not going to pretend I have a proven system for every homework type. But the principle transfers. Comprehension practice, vocabulary exercises, simple science experiment prompts calibrated to what a child already understands. These are all things a parent can describe specifically and get something useful back quickly. The constraint was never imagination. It was time.

What you already know how to do at work, you can do at home

If you have ever written a prompt for an AI tool at work, drafted a brief for a chatbot, or given feedback on an AI output to make it more useful, you already have the skills for this. The instincts are identical.

Specificity beats vagueness. “Make a worksheet” gives you something generic. “A4, three columns, 20 rows, multiplication from 2 to 9 by 2 to 12, name and score field at the top” gives you what you need. Start with a single page before committing to a full print run. Think about what your specific child actually needs, not what a generic child their age is supposed to need.

Most parents I know who are comfortable using AI at work have not yet made this connection at home. The tool is already open. The skills already exist. The gap is just the habit of reaching for it.

The shift underneath all of this

There is a version of parenting where you are a consumer of educational materials, working with what the market provides. And there is a version where you are a builder, describing exactly what your child needs and producing it in minutes.

I am not saying AI replaces good teaching or a parent’s attention. It does not. But it does change what a parent can reasonably produce with an hour and some thought. That feels worth paying attention to.

My daughters sat down with their sheets this week. Different names at the top, different questions throughout, no opportunity to copy. Small detail. But it came from actually knowing my kids, not from buying something designed for someone else’s.