Knowledge hub
Everything I publish, in one place: teardowns of real systems, analysis from inside the ecosystem and the Africa ↔ China perspective. Pick a lane.
Teardowns
Realms, PKCE, token clocks, brokering and the database nobody budgeted for. What actually breaks when single sign-on spans a dozen sites, and what to decide before it does.
Vector search retrieves what looks similar. It cannot tell you which shipped assemblies a bad plating bath touches two hops out. That is traversal, and traversal is a different operation.
The model is a component you swap. The harness around it, tool schemas, validation, sandboxes, budgets, permissions and audit logs, is where nearly every production AI incident actually starts.
Agentic tools have moved out of the private chat window and into the folders a team depends on. What that relocation actually changes, checked against vendor documentation as of August 2026.
The mark is not added to finished text. It is made of the model's word choices, and it can show a model was involved without ever showing how much of the writing was yours.
The model is the only part of an LLM application that cannot be tested deterministically. Everything around it can be, and that surrounding code is where most production incidents actually start.
Analysis
China's education ministry revoked or suspended 12,200 undergraduate programme slots in five years and added 10,200. The number everyone repeated is real. What it means has been reported wrong.
Generic reasoning costs $0.05 per million tokens. The premium moved to what nobody wrote down: how one mould behaves, what a Lagos trader really charges, what a supplier means by 差不多.
Token prices are set in dollars while African revenue is set in naira, and tokenizers charge some African languages nine times more tokens for the same sentence. That maths decides viability, not mode
An engineer who stays in Lagos and works for a foreign employer is present but economically absent. The country keeps the person and loses the tax base, the mentorship and the firm never built.
Hyperscaler capex reached roughly $745bn for 2026 while the two largest labs run at about $105bn combined. The bubble is real. The compute will outlive it. An engineer's read from Shenzhen.
The weights are real and the licences are real, and they say different things. Read the Chinese announcement next to the English one and the gift starts to look like a distribution strategy.
Signals
One claim, a few hundred words on what the coverage missed, and what would change my mind. Each is pinned to a primary source you can check.
Complete index
All 37 teardowns and 39 essays, newest first.