Madmaxlabs
Principles

How we build.

We've been writing software long enough to have opinions about it. Here are the ones we're committed to.

Care over speed

The manifesto says this longer. Shortened: care is what's missing from most software, and the thing that makes the difference isn't talent or capital. It's whether someone gives a shit about the thing being made.

We give a shit. The interactions, the defaults, the edges. All of it gets thought about. The bar isn't “this works.” It's “this is what you'd want it to be if you'd built it yourself.”

Useful over impressive

The current AI moment has filled the internet with software that demos well and fails in daily use. Slop. We avoid this by asking, on every feature, a single question: will the actual user actually benefit, or are we showing off?

Most things designed to look impressive don't help anyone. We'd rather ship the version that helps and looks boring than the version that demos great and frustrates.

Slow on purpose

Two people. No funding pressure. No quarterly metrics. We ship at the rate the work deserves. Sometimes that means a year between releases. Sometimes a flurry of changes in a week. Either way, nobody's rushing us to hit a number.

This shows up in the work. Edges get thought about. Bugs get fixed before features get added. Documentation exists.

Tools, not platforms

We're not trying to own you. The studio's apps are tools you use to do a thing, not platforms you live inside. When you use Ledgerly, you're not in “the Ledgerly ecosystem.” You're using a ledger. You can leave anytime, take your data, never come back.

The same will be true of every agentic app we ship. Bring your own key. Local-first. Export anytime. Tools.

Real people, real problems

Every project the studio takes on starts with someone we actually know who actually has the problem. Falcon's dad needed a way to share recipes. Falcon needed a focus tool with a sharp edge. We wanted a way to track personal spending without handing an opaque server our bank credentials.

That keeps the work honest. We can't bullshit ourselves about whether a thing is useful. Either the person we built it for uses it, or they don't.

Constraints we keep

The above translates into specific things we don't do:

  • No managed AI.BYOK on every agentic app we ship. Your conversations don't pass through us.
  • No engagement metrics in any dashboard. Time-on-app and daily-active-users get optimized for once you start watching them. The work drifts. We don't track them.
  • No dark patterns.The unsubscribe button is the same size and color as the subscribe button. Free trials don't auto-renew without a clear warning.
  • No third-party SDKs we don't need. Crash reports go to Google Play because the platform requires it. Otherwise the stack is us, the user, and the model.
  • No selling attention. If we can't charge for something directly, we don't ship it as ad-supported. We just don't ship it.

Care. Useful. Slow. Tools. Real. Constraints. The work that comes from this won't go viral. It might last.

The longer philosophy lives in the manifesto. The future-of-tech version lives at /agentic. Hold us to it.