AI, Strategy, and the Next Decade
Branching arguments, adaptive systems, the slow growth of a written body of work.
The next decade of AI is not a single global model. It is a constellation of national stacks, each tuned to its own language, its own data laws, its own definition of harm. The implications run deeper than most policymakers admit.
The thesis arrived in three parts: compute, data, and law. Each one is a sovereignty claim disguised as a technology decision. This essay walks through the three, then names the five governments that already understand it — and the dozen that don't yet.
Ep 014 · 54:12
v0.4 · public
Notebook coming soon
The next decade of leverage doesn't come from the model. It comes from what you write around it.
The next decade of AI is not a single global model. It is a constellation of national stacks, each tuned to its own language, its own data laws, its own definition of harm. The implications run deeper than most policymakers admit.
Training costs got the headlines. But the trillion-token inference workloads of the next decade rewrite the unit economics of every product touching an LLM. Here is the math that k…
I spent six weeks building autonomous agents that mostly failed. The interesting finding wasn't the failures — it was the substrate that emerged underneath them. Here's what to act…
Llama and Mistral didn't lose because their weights weren't good enough. They lost the application layer because nobody figured out the channels. A note on what wins next.
with Anonymous — EU AI Act drafting team
A long conversation about why every major economy is suddenly racing to build a national AI stack — and the policy levers that will decide who actually pulls it off.
A reference architecture for retrieval-grounded agents that actually ship to production. Includes the eval harness, the routing layer, and the parts I would not build again.
A map of the AI stack, updated monthly. Drag-and-drop layers, costed in real numbers, exportable as a Mermaid diagram.
A multi-model debate harness. Six models argue a question, a seventh judges. Used as a research substrate for a few of the essays here.
A small writing tool I use to draft these essays. Distraction-free, obsidian-friendly, with an inline LLM that argues back.
I'm Rijul. I write essays, host a podcast, and build small things on the web — all of it in service of one question: how do we leverage AI in the next decade without giving away what mattered in the last? New work lands here when it's ready. Subscribe and I'll send it once.