Writing
Writing
Most of what I write goes into The Intelligence Estate, a weekly first-person series on company knowledge, AI memory, and the systems built on top of them. I write it while the work is in progress, so it includes the parts that didn’t work.
Below is a selection. Each link goes to the original on Elevare’s site.
-
The Most Important Thing I Built This Week Was a Boundary
Connecting every source gives software more context, not better judgment. On packing for the decision in front of you, and why restraint becomes an advantage when adding features costs almost nothing.
-
The Score Isn’t the Intelligence
An 87 for what? Why every score hides an assumption, why scoring is closer to policy than math, and the questions to ask before trusting a number.
-
The Next Move Belongs to the Market
On telling an internal unknown from an external one. Once the important questions can only be answered by a customer, more building stops being progress.
-
Weekly Knowledge Sweeps
Capturing everything is the easy part. The weekly sweep decides what survives, what becomes a task, and what should disappear.
-
The Model Should Be Replaceable
If your system stops working when one AI tool goes away, you rented a brain. Why institutional knowledge has to live outside the model.
-
Every Conversation Starts Over
The bottleneck with AI is rarely the model. It’s that every conversation starts from zero, and the fix is a memory the work can accumulate in.
-
The Intelligence Estate
The opening entry. Knowledge rarely disappears because nobody wrote it down; it disappears because it was written down in forty places. What it takes to give it one home that people and AI can both use.
The full series has entries not listed here. New entries also appear on LinkedIn.