Why we build our own products first

We don't ask clients to trust a method we haven't used ourselves.

Anyone can describe a process for building with AI. It's a different thing to run that process on your own products, with your own money and time on the line, and live with the results. So that's what we do first. Our own products are the proof.

Varsity Page and VP Core

Varsity Page and VP Core together make a two-sided school spirit platform that turns high schools into sponsorship revenue engines and puts every program in the spotlight, from the marching band to the robotics team, not just the varsity football team.

I built it by myself in nine months.

That's only possible because of how it was built. I work as what I call a PMgineer: one person doing the product manager, AI engineer, software engineer, DevOps and forward-deployed engineer work, directing a fleet of AI agents. I talk more about what that role means in my takeaways from the Product Insider podcast.

Directing agents only works if they know what they're building and why. Every part of Varsity Page started with a written why: who it's for, what problem it solves, what it must never do. Varsity Page handles information about students, so the privacy rules aren't a feature request. They're written down before any agent touches the code, and every agent building the product has to follow them.

lmn

lmn is our framework for directing AI coding agents from written specs and guardrails.

AI is a capable but naive assistant. It builds exactly what you describe and misses everything you didn't. lmn exists to make "what you describe" complete: every task starts as a written spec, every agent works inside written rules, and every change gets checked against both before it ships.

It's how we ship fast without shipping chaos. The speed comes from the agents. The control comes from the specs and guardrails.

Filament

Filament is a desktop app for managing projects and credentials across AI-driven development. It's in development.

It came out of a real need from building the other products: when one person is directing many agents across many projects, keeping track of which project uses what, and keeping secrets where they belong, becomes its own job. Filament is our answer to that, and it's being built the same way as everything else, starting from a written why.

What building our own products teaches us

Each product started with a written why. Each one taught us something we now bring to client work:

• The why has to be written down, not just agreed on. People can fill in gaps from a conversation. Agents can't. A why that only lives in someone's head doesn't reach the AI doing the work.
• Governance is a product decision. Building a product that handles student information made that concrete. The rules came first, not after a problem forced them.
• Process before AI applies to us too. The same order I describe in Stop starting with AI, name the real problem, build a working process, then add AI, is the order we use on our own work.
• Speed without structure is just faster mistakes. AI agents will build the wrong thing quickly and well if you let them. Specs and guardrails are what make the speed useful.

This is also why I start every client project with the same three questions before any AI build. We learned the hard way, on our own products, what happens when one of them is skipped.

Building yours the same way

When you work with luminAIt, you get the method we use on our own products, not a theory about one. The same written why, the same order of operations, the same governance before scale.

If you have a product idea and want it built the same way, book a free consultation. You can also see more of what we've built on our products page.