The Fifteen-Year Project
Raising him on purpose, starting with a document we haven't written yet
My wife and I are about to write a document we’re calling our Values Constitution. The question it forces is simple and uncomfortable: what values do we actually want to be the foundation of how our son is raised?
Almost no family has that document. Not because they don’t have values - every family does - but because nothing ever forces the question. Values stay implicit. And implicit values get overridden in the moment by whatever’s convenient, whatever the culture is nudging toward, whatever the AI tool defaults to.
The Constitution is the thing that turns implicit values into something deliberate. Something you can hand to a tool and say: these are the boundaries. These are non-negotiable. When your defaults conflict with these values, these values win.
Writing it is the work we’re about to start. It won’t be fast. It won’t be easy. Most of it will probably be the uncomfortable parts - the domains where we both thought we agreed and will discover we don’t. That’s the point. Better to find those gaps now, on paper, deliberately, than to discover them ten years from now in our son’s behavior.
That document becomes the foundation of everything else. No tool gets built on top of it until it exists. No AI gets configured without deferring to it. No material gets delivered to Dominic that hasn’t been filtered through it.
The values come first. Everything else is implementation.
How we got here
I’m an entrepreneur. I have an eight-year-old son, Dominic, and my wife and I are homeschooling him.
For a while, I shared the foundational work with my wife - the math, the English, the science. At some point we sat down and asked a different question: what is the most valuable use of the limited time I actually have with him?
The answer we arrived at, together, was that the foundational work was covered. My wife handles it, and handles it well. What wasn’t covered was the other thing - the mindset, the entrepreneurial way of thinking, the part of his formation that I’m uniquely positioned to give him because it’s what I’ve spent my adult life doing.
So we made a decision. She’d take the foundational subjects. I’d take the one day a week I have with Dom and dedicate it to the work that only I can do with him. Not homework help. Not supplementary tutoring. Mindset. How to see problems. How to build. How to think like someone who creates value rather than someone who waits to receive it.
That decision is the reason I started thinking about everything else I’m about to describe.
Because once you decide that the time you have with your kid is going to be spent on something specific and irreplaceable, you start looking at every tool, every platform, every outside influence through a different lens. You start asking whether what’s entering his formation is helping or competing with what you’re trying to build.
The AI problem nobody talks about
Pretty early on, I assumed I’d eventually use AI to help with this. Entrepreneurs use it. Kids will grow up with it. Why wouldn’t I use it to help Dominic build the mindset I want him to have?
When I actually sat with that question, something shifted.
Every AI tool has values baked into it by its creators - the subtle ones. The ones that shape which historical figures get framed as heroes and which get framed as complicated. The ones that decide what counts as success in a story. The ones that determine whether effort, grit, and discipline get emphasized or softened. The ones that choose what framings of family, faith, authority, tradition, gender, work, and meaning get treated as default.
These aren’t neutral. They reflect the worldview of the companies and teams that trained the models. That’s fine for generic tasks. It’s not fine when the tool is going to shape a child for years.
Here’s the part that caught me: if I was already being deliberate about how my limited time with Dom got used, how could I not be deliberate about what other influences were shaping him through whatever tools we eventually introduced? My one day a week is precious because it’s specific and chosen. The moment I bring in an AI tool without the same deliberation, I’m letting someone else’s unexamined defaults contribute to his formation alongside my carefully chosen input.
The tool that waits
There’s a tension at the center of this project, and I want to name it rather than pretend it isn’t there.
Dominic is eight. He’s growing up in a world where kids his age are already on devices, already interacting with AI, already surrounded by screens. And here I am, building an AI tool in service of raising him.
The resolution: the tool isn’t for him yet. Not directly. The AI I’m building is a support tool for my wife and me. It helps her plan foundational lessons. It helps me prepare the entrepreneurial day. It captures what happens during our time with him so we have a record of who he’s becoming over years, not just a vague parental memory. Dominic doesn’t interact with it.
Eventually - when he’s older, when we decide together the time is right — the tool will be available to him. By then it will have years of context about who he is, what he’s interested in, what he’s struggled with, what he’s loved. It won’t be a generic AI meeting a teenager. It’ll be something that has been watching him grow, shaped by the values his parents deliberately encoded into it.
That’s the long arc. The short arc is that right now, this is a parent tool. A co-parenting tool. Not a kid tool. And by the time it becomes his, it will have been shaped for years by the specific values we chose for him, rather than the generic defaults that would shape him if we let any off-the-shelf AI do the job.
Why this has to be a fifteen-year project
Most educational tools are designed for a semester. Maybe a year. The system I’m building is designed for fifteen years - from Dominic at eight to Dominic at twenty-three and beyond.
That changes everything about how it’s structured.
A semester tool can afford to be vendor-locked, platform-dependent, format-proprietary. A fifteen-year tool can’t. Whatever infrastructure exists today probably won’t exist in the same form in 2040. So the tool has to be built on formats that survive - plain text, simple files, structures any future system can read. The memory of who my son is can’t be trapped inside a database owned by a company that might not exist in ten years.
It also means the tool has to grow. What my son needs at eight is completely different from what he’ll need at fifteen or at twenty-two. The architecture has to accommodate every version of him across that arc. A tool built for who he is today becomes useless the moment he changes.
The other thing a fifteen-year horizon forces is a kind of patience I’m not used to as an entrepreneur. There’s no quarter. There’s no launch. There’s no metric that captures whether this is working in the way business metrics work. The only honest measurement is: who is Dominic at twenty-three, and how much of that came from deliberate formation versus drift?
There’s another implication worth naming directly: if this architecture is sound, if the values encoding holds, if the tool genuinely helps raise him the way we want him raised, then what I’m building for my family is the same thing other entrepreneurial families would want for theirs. I’m not designing it as a product right now. I’m designing it for my son. But I’m designing it in a way that a product could eventually emerge from, if what we learn over these years turns out to be worth offering to other parents.
I’m not selling anything yet. I may never. But I’m building it cleanly enough that the option stays open.
What we’re actually building
Here’s the shape of it, without the technical details.
There’s a framework - a universal layer - that any family could theoretically use. It contains the structure: how values get encoded, how parent voices get captured, how a child’s profile gets maintained over time, how the tool learns to sound like Mom or Dad rather than like a generic AI, how materials get generated and reviewed before anything reaches the child.
And there’s a family layer - our layer - that’s private. It contains our actual values (once we write them), our actual voices, our actual profile of Dominic, everything specific to us. That layer doesn’t get shared with anyone. Not now. Not ever. It’s the soul of the system and it belongs to us.
The separation matters for two reasons.
One, it protects our privacy. Anyone who ever saw the framework could learn how the system works without seeing a single word about our family.
Two, it makes the framework potentially useful to others down the road. If what we build works, the framework can be offered to other families who plug in their own values, their own voices, their own children. Same architecture. Entirely different souls. Every family gets a tool shaped by their own deliberate decisions rather than by defaults someone else chose.
That second part is a long way off. I’m not in a hurry. But I want to build it right from day one because retrofitting privacy and portability later is painful, and starting correctly is nearly free.
Why this fits with DAC
If you’ve been reading The DAC Life, you know I write about how decisions actually get made. The short version: most outcomes fail not because people make the wrong decisions, but because commitment gets requested before the decision to change has been made. The sequence is wrong. The decision has to come first.
I’ve written about this in the context of sales, marketing, funnels, and offers. I’m realizing it applies everywhere - including the most important decisions most of us ever make, which are about how we raise our kids and what we want to build.
Most parenting runs the broken sequence. Commitments get made - to schools, to schedules, to curriculums, to peer groups - before the decision has been made about what we’re actually trying to create. The awareness comes later, usually too late, when the outcomes don’t match what the parents thought they were doing.
Reversing that sequence, deliberately, inside my own family, is the thing I’m actually testing.
The Values Constitution is the decision. The AI tool, the homeschooling infrastructure, the day with Dom, the long arc - all of it is downstream of the decision. If the decision is right, the implementation almost builds itself. If the decision is wrong, or worse, un-made, no amount of beautiful implementation will fix it.
DAC in business was the practice. DAC at home is the test. If the framework holds up here - in the highest-stakes, longest-duration application I can run - then I know it’s real. If it doesn’t, I’ll learn that too.
What this series will and won’t be
This isn’t a regular column. It’s occasional field notes from a live project. I’ll write when there’s something real to report. If the Constitution takes three months to draft, the next post comes in three months, not next week.
I’m not selling anything. There may be something to offer someday, but right now this is a passion project documented publicly because I think the reasoning might be useful to other families thinking through similar questions.
I won’t share our Values Constitution when it’s done. That’s private. I’ll share the process, the shape of it, what it revealed, what it cost to write. The artifact is ours. The thinking is worth making visible.
And I won’t pretend this is further along than it is. Right now I’m at the threshold of the work. The Constitution hasn’t been written. The infrastructure exists but isn’t populated. The first real session with the tool supporting a day with Dom is ahead of me, not behind me. I’ll report from where I actually am.
What’s next
My wife and I are going to sit down and start the Constitution. It will probably be messy. We’ll disagree on things we didn’t know we disagreed on. We’ll probably take longer than we expect. When there’s something real to say about what that process revealed, I’ll write about it.
In the meantime, I’m finishing the infrastructure that will sit underneath everything - the support tools for my wife and me as we homeschool him, the memory system that will grow with him, the framework that eventually becomes his.
If you’re thinking about similar questions for your own family, or if the idea of deliberate formation resonates, follow along. This is going to be a slower newsletter than most. The stakes are high enough that I’d rather get it right than get it frequent.
More when there’s more.
-- Razvan