One of the most valuable lessons I have learned in personal and professional life is that it's OK to take an idea you are initially thrilled about and "put it on ice." Maybe the timing isn't right, or maybe the idea has not fully grown into its potential. By placing it on hold, you don't suffer the negative feeling of wasted energy or second-guessing your ability to generate novel concepts.

A classic example of this was my (fairly) recent attempt at writing a novel. Loved the world building, the character creation, the plot conceptualization, and the attempt at a great underlying theme. However, I strangely felt guilty putting significant time into it because of so many other pressing life priorities.

So what did I do? Put it on ice. Or maybe in the closet. An icy closet.

Fast forward a year later, and a new need was brewing.

Every summer, towards the back half, I try to get my girls back into learning mode to ease the transition back to school. Last summer, Prodigy was a huge hit. However, they have basically beaten the game and the learning interest has tapered off. As I searched around, I found lots of interesting games, but nothing that would really hold their attention for significant time. The alternative, printing worksheets and planning a learning schedule, was like a UN summit on how to distribute the world's wealth evenly.

Then it hit me. Why couldn't you wrap math and writing practice into the kind of immersive storyline you find in movies and games? It's surely been tried, but there's no "Harry Potter of finishing a grade level" yet. The best part: it doesn't take anything away from how teachers and tutors help students learn. The story is only meant to incentivize and shape some examples of the learning. Every student has different learning needs.

So I started to think: what is the limiting factor here? The obvious answer, an engaging story. You can probably see where this is going :)

I decided to pull the epic idea out of the icebox and see if I could use parts of it to create an engaging summer program. Turns out, with a lot of coffee and a commitment to finishing, the pieces started coming together, and Sylverlight Mysteries was born.

One story, two kids

Here's the design decision I cared about most. My girls are heading into 2nd and 4th grade, which normally means two completely separate sets of material and zero overlap in what they're doing. Sylverlight is one story with two difficulty tracks. Same island, same characters, same mysterious guide, same villain. Only the problem sets are different.

That was the whole point from day one. My 4th grader is borrowing across zeros while my 2nd grader is regrouping within 100, but they both caught the same doctored tally board, and they can argue about the hooded villain at dinner. Learning as a shared world instead of two separate chores.

The structure is simple: four mysteries, each one solved with exactly the math and reading skills they need for the coming year. And the math is not bolted onto the story. The math IS how the mystery gets solved. You catch the thief by noticing the numbers on the tally board don't match the crates, or that the wood is shorter than the delivery slip claims. The whole series quietly teaches one habit: don't trust a number just because it looks about right. Check it yourself.

Yes, I used AI. Here's the honest version.

I know the question is coming, so let's get into it. I built this heavily with generative AI, and it was also part of the motivation. At work I use AI constantly, but almost entirely for coding and technical tasks. I wanted to see how far the recent models could go on something creative.

The best way I can describe the experience: the AI was the editor who could also do the heavy lifting on prose. It drafted scenes in a voice I could never sustain across a hundred pages. It kept track of curriculum standards while I thought about story. When I needed a name for a plump, dumb, tasty fish, it gave me ten options so I could find the right one. (Hucklefish, obviously.)

But it was not the author. I was. Every call that mattered was mine: the villain's methods, what my kids would find scary versus thrilling, which lines sounded like my world and which didn't. I caught the continuity errors. I softened the line that was too dark for a 7-year-old. I rejected entire directions. If you have ever worked with a talented collaborator who has no real stake in the outcome, you know the job: the taste, the stakes, and the final say stay with you.

And for anyone concerned that an AI helped write my kids' summer prep: nobody remembers every plotline from the cartoons they grew up on, and nobody needs to remember Sylverlight either. The story is not the curriculum. The story is the reason to DO the curriculum. If a detail is ever wrong, it's fixable. The math has an answer key.

The part I didn't plan

The villain of Sylverlight never steals big. He steals a little at a time, from everyone, in amounts too small for anyone to notice, and he counts on nobody looking closely. The heroes are simply the people willing to look closely anyway.

I won't pretend I set out to write some grand message. But I'll admit the theme rhymes with things I know. Summer learning doesn't vanish all at once; it slips a little at a time. Ideas don't get abandoned in one dramatic moment; they get shelved once, then again, then again, in increments too small to feel like a decision.

And here's the part I genuinely did not notice until it was pointed out to me. In the story's finale, the kid has been avoiding the call to go back. The magical light that always appeared for them has stopped coming. And when they finally decide they're ready to face what they've been avoiding, the doorway that opens for them... is a closet.

I wrote my own icy closet into the climax of the story without realizing it. Which I suppose is the whole point of idea resurrection: the ideas you put on ice don't die in there. They wait. The novel even donated its DNA to Sylverlight, since the island survived, and so did the villain, who in the original epic answers to something much larger. If this experiment works, that's where we go next.

Tuesday, we find out

The plan is deliberately light, because I still believe summer should be mostly play and family. Two sessions a week: math on Tuesday, writing on Wednesday, eight sessions total across four weeks.

I gave myself 30 days to build it, starting June 14th, and finished a couple of days early. Now comes the part no AI can help with. I built a villain, a talking fox, a founding myth, and an entire island for an audience of exactly two, and there is a real chance they take one look and see straight through to the worksheet underneath. Kids have an unerring nose for vegetables hidden in the pasta sauce.

Maybe they'll shrug. Maybe they'll fight it. Or maybe, when the story asks them to check the numbers on the tally board because a river is counting on them, they'll lean in.

We start the week of July 20th. Then we find out.