Ctrl + Cmd AI

We start with an idea—or just a thought, really.
Then we observe, theorize, test, document what happens, and loop back to do it all over again.

Rinse, repeat.

Below is basically years of that cycle—ideas, attempts, mostly failures, but always learning something new. Here's us trying to actually show you what that looks like.

The runs — one experiment, newest first

A supplement store was charging the wrong amount when someone bought more than two of a thing. Annoying — probably a rounding bug, the kind you fix in ten minutes and forget.

It was not a rounding bug. The store was letting the customer's own computer decide the price. We set a fifty-seven dollar order to two thousand and the store said yes. Run it the other way and fifty dollars of product costs two.

The fix was the obvious one. What wasn't obvious is that our fix, before it ever shipped, would have thrown out real customers' orders — and we only found that because somebody other than the person who wrote it went looking.

▸ Result

You don't trust your own fix. The bug you can see is never the one that gets you; the one that gets you is sitting inside the thing you were sure about. The person who builds the fix and the thing that checks the fix are never the same.

The log

New experiments go out as they finish.

What we tried, what it cost, what actually happened. No pitch, and no schedule it has to keep.