Two people can walk into the same situation with the same problem and walk out with completely different outcomes — because one of them happened to get the better expert. Kernwit is about closing that gap.
The problem
Every one of these relationships has the same shape. One side holds knowledge. The other side needs it, and has no reliable way to judge what they're getting.
You can't evaluate a diagnosis without medical training. You can't tell good legal advice from bad until it's too late to matter. So the quality of what you receive comes down to the luck of the draw — and the people with least room to absorb a bad outcome usually get the worst draw.
The idea, and its timing
The premise was that expert judgment could be made more consistent — that the floor could be raised, so the advice you get depends less on who is giving it. At the time that meant trying to encode expertise by hand, domain by domain. It didn't work, and we stopped.
The problem was never the idea. It was that the tools to make expertise legible didn't exist yet.
Large language models changed which half of the problem is hard. Reading a case, comparing it against what's known, and explaining the reasoning in plain language used to be the research problem. Now it's mostly engineering — and what's left is the part that always mattered: judgment about where a machine should and shouldn't be trusted, and being honest with people about the difference.
In the meantime
Rutino is a habit app about getting better at one thing — built with, and mostly designed by, a 14-year-old. No ads, no trackers, and your data stays on your phone.
It's a small idea from the same place as the big one: that the useful thing is usually raising the floor rather than optimising the ceiling. Most habit apps are built for people who were already going to succeed. Rutino is built for the week you nearly quit.