I've built 3 products in six years.

It took me until this year to notice they were all the same company.

A synthetic monitoring platform that existed because enterprises refused to take their own dashboards at their word.

We verified from the outside, like a real user.

A prayer app where the AI is never allowed to play the oracle.

It listens, then routes to Scripture. The trusted source gives the answer. The software is the servant, never the source.

An open-source framework that grades AI agents on the path they took, not just the final answer, and refuses to emit a score without an explanation.

One thesis underneath all of it. Don't ask people to trust you, hand them a way to verify.

I think that thesis is about to become the whole game.

When generating a convincing anything costs nothing, proof is the only currency that doesn't inflate.

I wrote the full argument, including my prediction for where the next decade of value goes.

Full article in the comments.