know.2nth.ai is a deep technical knowledge tree — built for engineers and the agents that read it. These briefings are its other half: the same subjects, stripped of jargon and told as the decisions a leader actually has to make. Short, visual, honest about the trade-offs — and each one links to the full technical version for the people who'll build it.
Each briefing is a quick read. Every technical term is translated to a business one, with the original noted for your team.
Framed around cost, risk, ownership and leverage — the things a CEO weighs — not feature lists or architecture diagrams.
Every claim sits on a full technical leaf in the knowledge tree. Hand the deep version to whoever has to make it true.
Marc Andreessen says programmer, designer and PM are collapsing into one “builder” — the same person we call the operator, commanding AI agents with human specialists in the loop. What the convergence means for your org chart.
Open the briefing →Rent on someone else's tech, a services margin to run it, and an FX line because it's in dollars — and none of that is malice, it's the incentive. Why renting was rational until agents changed the cost of owning, and the honest barbell for taking your stack back.
The software, settings, customisations and data you run on are real assets. This is what it means to actually own them — and the lock-in risk when you don't.
Replace expensive software gradually — one function at a time, behind a façade, with the old system as a safety net — using open source, owned code and AI.
Edge-first computing as border control for your data — the checkpoint where you decide what stays in SA and what may cross. The plain answer to POPIA and AI residency.
An agent is a brilliant new hire with zero common sense. The ladder of autonomy, the safe zone vs the danger zone, the guardrails that make it safe, and where agents go wrong.
Connect your Shopify storefront to an owned back office and finally get real stock, true cost and margin per product, and buying that follows your sales. For makers with many suppliers.
An agent is only as good as the data it can reach. The bottleneck moves to scoping and clean, reachable data — fix those first or you buy faster chaos. Based on the Ng × Chase thesis.
Most things sold as AI agents should be simpler, cheaper, and more predictable: a fixed workflow, not an autonomous one. The difference, the ladder from prompt to agent, and four questions before you greenlight one.
Because the model was never the expensive part — the scaffolding around it is. Where the cheap model wins, why switching is a rebuild not a swap, who should move now, and why a backstop you control is the real decision.
A borrowed lens (Kant), kept concrete: why learned procedures beat hard-coded rules, why your data’s value is in its synthesis rules not the pile, and why mature leaders plan “as-if” while knowing the exact boundaries of the tool.
Because the queue moved: from writing code to deciding, specifying and reviewing. Why your velocity numbers will look strange, three questions for your engineering lead, and why the governance fix is boring on purpose.
It's called Dataverse: the governed store your Microsoft apps and Copilot agents already run inside. What it is in plain terms, why it keeps data in South Africa, how it sits under Copilot, and where the lock-in bites.
It's five decisions, not one, and you only need two: own the router and a small model on your own data, rent the frontier model that's obsolete by Christmas. The three reasons to build in-house, translated — and why a backstop you control is the real question.
The modern publishing stack told as one story for non-engineers: who does what, the push-to-live pipeline, near-zero hosting, marketing autonomy — and the honest lock-in and when to skip it.
The front office is the tip — website, Shopify, portal. The back office is the bigger mass below: ERP, Git, suppliers, finance, HR — where cost, risk and your IP really sit. Rent the front, own the back.
We're turning the heaviest, most technical corners of the knowledge tree into briefings a leader can read in one coffee. Candidates on the list:
If there's a technology call your board is wrestling with — a renewal, a migration, an AI question — it can become the next briefing. The knowledge tree almost certainly already has the technical depth; a briefing just turns it into the version you can put in front of the people who decide.