# Wired: Deep Briefings and Ideal Setups > [!summary] The shortest account > **Wired** is the vault's **briefing layer**: wired about anything interesting, with a standing bias toward **artificial intelligence**. Gadgets, service providers, new medical treatments, local models, data centers, transformers — each entry tells you what the thing actually is and the ideal way to handle it. No argument, no interpretation: the briefing and the setup. > [!info] How this differs from [[collections/welcome|Collections]] > Collections organize what Bryant is presently investigating, interpreting, and connecting — argument-bearing maps of larger systems. Articles make bounded arguments. **Wired makes no argument at all.** A wired entry answers two questions and only two: *what is this actually*, and *what is the ideal way to handle it*. It is reference material with briefings and prescriptions — what is real, what it costs, what to get, what to avoid — not interpretation. ## Start here The section opens with five founding entries on AI — the standing bias — and stays open to anything interesting after that. Each entry answers the same two questions: *what is this actually*, and *what is the ideal way to handle it*. - **What is contemporary AI, technically — and what is the reference configuration for working with it as a builder?** → **Wired About AI** - **What does it take to run capable models on your own hardware — and what is the ideal local-model workstation, specified end to end?** → **Wired About Your Own Local Model** - **How do you get from a local model to a local agent — and what is the ideal personal-agent stack on local iron?** → **Wired About Building a World-Class Local Bot** - **What does hyperscale AI infrastructure actually look like — and how do you read a data center the way its architects do?** → **Wired About the Inside of Data Centers** - **What is the transformer architecture, from the metal up — and what is the ideal tooling for inspecting and reasoning about transformer behavior yourself?** → **Wired About Understanding Transformers** ## What a wired entry does A wired entry performs two jobs, in order. First, the **briefing**: what the thing actually is — how it works, who the serious players are, what is real versus marketing, what changed. Second, the **ideal setup**: the recommended way to engage it — what to get, who to use, what it costs, the steps in order, the tradeoffs at each decision point. Numbers where numbers exist. The register is plain, grounded, and decisive. This is the section you read before you buy, build, book, or commit. ## Launch index | Entry | Slug | Covers | Status | |---|---|---|---| | Wired About AI | `wired-about-ai` | What contemporary AI systems are, how they are trained and served, and where the capability lives in the stack | Queued | | Wired About Your Own Local Model | `wired-about-your-own-local-model` | Quantization, runtimes, VRAM math, and the frontier of what fits on a desk | Queued | | Wired About Building a World-Class Local Bot | `wired-about-building-a-world-class-local-bot` | Tool use, memory, retrieval, scheduling — the harness that turns inference into assistance | Queued | | Wired About the Inside of Data Centers | `wired-about-the-inside-of-data-centers` | Power, cooling, networking fabric, GPU clusters, and the physical economics of a training run | Queued | | Wired About Understanding Transformers | `wired-about-understanding-transformers` | Attention, tokenization, training dynamics, and inference mechanics, from the metal up | Queued | ## What "wired" means here Wired about a thing means knowing it the way an electrician knows a panel: current paths, load limits, failure modes, and exactly which wire to move to change the behavior — whether the panel is a transformer architecture, a gadget, a clinic, or a cloud provider. This section exists so that nothing in the corpus floats free of the reality it describes.