The Guide · Working Knowledge

Short routes

Or choose a reading path.

If you read the newsMake sense of the headlines · 3 stops If you want to buildDo the work yourself · 5 stops If you're picking toolsCompare agents by harness, not hype · 2 stops The quick tourA short orientation · 3 stops If you're non-technicalSkip the engineering deep dives · 5 stops If you're investing / execMoney and the board · 3 stops

What AI Actually Is (and Isn't)

AI, machine learning, deep learning, and generative AI overlap, but they are not synonyms. Knowing the layers makes the headlines much easier to read.

3 min read →02

How a Model Actually “Thinks”

A language model produces one token at a time from a probability distribution. Sampling, context, tools, and training turn that simple interface into surprisingly complex behavior.

3 min read →03

What's an API—and Why It's the On-Ramp for AI

An API is a defined contract between programs. It is how applications call hosted models and how tool-using systems reach other software.

3 min read →04

How to Get Useful Answers (and Know When Not to Trust Them)

Good prompting is mostly good briefing: state the goal, provide the right evidence, define the output, and specify how the result will be checked.

3 min read →05

What You Can Actually Do With AI

Most useful products combine a small set of patterns: assist, retrieve, act, code, generate media, and structure information.

3 min read →06

What's an Agent—and Why They're Hard

An agent wraps a model in a loop with tools, state, and stopping rules. The capability is real; reliability and safe access are system-design problems.

3 min read →07

Follow the Money

AI economics link training, inference, chips, memory, networking, data centers, and power. The headline numbers are estimates and forecasts, so label them accordingly.

3 min read →08

Working in the Terminal

A coding agent uses ordinary files, shell commands, Git, dependencies, tests, and deployment tools. Reading that loop helps you supervise it safely.

3 min read →