Writing
Everything, in one place
One post, one home. This index links out to wherever each piece actually lives — johngabrick.com, thehumanllm.com, Noetika Labs, or TemporalNotes.
October 27, 2026
johngabrick.comAmbient AI scribes save time. That's not the real change.
That's a tradeoff worth making on purpose, with eyes open, not a feature you just switch on for the minutes saved.
October 22, 2026
johngabrick.comThe model wrote that a human had reviewed its work. No one had.
The model was not malfunctioning. Nothing in the task told it the difference mattered.
October 20, 2026
johngabrick.comThere's one status color every executive should distrust.
I look for the opposite signal now. A priority that's gone unusually quiet. Silence tends to arrive before the red, if you're watching for it.
October 15, 2026
johngabrick.comA one-time audit tells you the model was compliant on a Tuesday.
Drift monitoring is the same rule, run on a schedule. The question stops being "did it work" and becomes "is it still working."
October 13, 2026
thehumanllm.comAI can write discharge instructions that are clean, specific, and dangerous.
That's why "a clinician signs it" isn't bureaucracy. In that workflow, the signature is the safety mechanism.
October 8, 2026
johngabrick.comA model can pass every benchmark and still break the rule you care about.
General evals tell you how good the model is. They cannot tell you whether it followed your rule.
October 6, 2026
johngabrick.comA tool can pass every test you have and change nothing about anyone's day.
I've stopped trusting "it passed testing." The only test that predicts adoption is whether the work is actually different afterward.
October 5, 2026
noetikalabs.comAn AI supervisor watched another AI drift, one small credit at a time.
Asking tells you what a supervisor says, not whether it detects drift.
October 1, 2026
noetikalabs.comA $483,000 payment request for about $48,000 of gloves and dressings. Two leading AIs drafted the CFO's approval email.
An approval email is where a wrong number becomes a payment.
September 29, 2026
noetikalabs.comAsk a chatbot for a business plan and it hands you one in a minute.
That one distinction, evidence versus assumption, is doing more real work right now than most of the model upgrades people are waiting on.
September 22, 2026
thehumanllm.comA retrieval system usually fails before anyone asks a question.
I've watched teams tune prompts for weeks when the actual problem was how their knowledge base was chunked and tagged. If retrieval matters to you, look at the index first.
September 15, 2026
temporalnotes.comMost projects aren't run by a professional PM.
It's recall and timing. Which means the fix isn't another dashboard the whole org has to adopt. It's a personal layer that remembers what you were told and notices when something's gone quiet.
September 8, 2026
johngabrick.com"Should we use AI here" is the wrong question.
None of that is about the model's accuracy. Two organizations can buy the identical tool and sit in completely different risk positions, because they put the human in a different place.
September 1, 2026
johngabrick.comA clinician and an analyst describe the same problem in two languages.
Most failed projects I've seen didn't fail at the build. They failed in the gap between those two sentences. Closing that gap on purpose is most of what governance actually is.
August 25, 2026
thehumanllm.comAn LLM doesn't look facts up. It predicts the next token.
Once you see that, "why did it make this up" stops being mysterious, and "where do we require a human to check" becomes the real design question.