The risk is what you don't see
On unknown unknowns, inherited assumptions, and why good process makes uncertainty harder to ignore.
The problem often isn't what you don't know. It's what you think you know that ain't so.
Sometime around 2011 or 2012, my physics lecturer said something like that. At the time, I remember thinking: Ohhh, deep. Also, he is a poet and he doesn't know it.
But the line stuck with me. A lot of my career since then has been about unknown unknowns—especially in safety-oriented AI development for autonomous driving.
Ignorance is bliss
And the uncomfortable truth is, ignorance is bliss.
I have seen too many “it's probably good enough” decisions made without anyone really knowing whether we had even thought of the right things to test. Then the unknowns get passed downstream.
Maybe QA catches something. Maybe a customer does. Maybe a security issue gets exposed. Maybe the incident becomes one of those brand-killing “oops” moments. So we blame the technology.
But was the technology really the whole problem?
Structure helps us see
When we build bridges, ships, homes, and cars, we do not rely only on individual brilliance or heroic last-minute checks.
We rely on structure. Process. Risk awareness. Assumptions made visible. Failure modes explored before they become expensive.
So why should AI-enabled software, or software built with autonomous tools, be different?
Pro tip: it shouldn't be.
Why I am building Agentic Blue
That is why I am building Agentic Blue together with two similarly obsessed people. Not because process magically prevents mistakes. It doesn't.
But good process helps you notice what you would otherwise miss. It helps teams ask better questions earlier. It reduces the chance that two of the 50 obvious and sometimes embarrassing things only become obvious after release.
That is the point for me. Help with the process. Lean on the process. Refine the process.
Not to eliminate uncertainty, but to make uncertainty harder to ignore.
If you want to hear more, write me.