Skip to content
INTELIGÊNCIA ANALÓGICABlogSign in
ESSAYS FROM THE HOUSE

What I'm thinking about

One idea, applied to different things. Essays that stand on their own — and that, together, say the same thing: the more the machine settles, the heavier what is left for a person to decide.

#01 02 Pseudo-process: it exists in theory, not in practice Checking the sign is not checking the room READ ▸ #02 02 Responsibility does not shrink with competence The better AI gets, the more someone has to answer for it READ ▸ #03 02 Producing got cheap. Answering for it did not. Your company now runs on two currencies READ ▸ #04 03 The loss that never becomes an incident is the one nobody measures. Half your team uses AI in secret. The other half doesn't even try. READ ▸ #05 02 The tool answers "how". Nobody sells you the answers to "what" and "who". Buying the tool is not having governance READ ▸ #06 02 The bill is already open. What's missing is the paper showing you knew. You don't need a new AI law. The duties already apply. READ ▸ #07 03 Anyone promising it closes is selling what they cannot deliver. No AI control closes 100% READ ▸ #08 03 Automating what nobody understands produces a system nobody can check. AI's easy promise is to replace. The one it keeps is to inform. READ ▸ #09 02 Technology amplifies the organisation that exists — or its absence. Every company is becoming a technology company. Most of them drown. READ ▸ #10 02 Nobody builds autonomy on top of chaos. Order comes first. The four steps: where your company stands READ ▸ #11 03 If you cannot see what is happening, all you can do is forbid. Visibility is what lets you allow. Blocking AI is not knowing what it is doing READ ▸ #12 02 What comes in outside the queue is not urgent. It is invisible. Priority is impact, not rank READ ▸ #13 02 Switching on impulse costs more than the mistake that prompted it. A supplier is not dropped for making a mistake. They are dropped for repeating it. READ ▸ #14 02 You buy the world's way. You build your company's way. Build or buy: is it your company's way, or the world's way? READ ▸ #15 02 You write down what is a risk to leave unspoken. The rest is decoration. You do not need to document everything. You need a yardstick. READ ▸ #16 02 Whoever answers for the business has the right to decide about AI with judgement — without becoming technical. Clarity is not a privilege of those who understand technology READ ▸ #17 02 Autonomy does not make its debut in the process that stops the company if it fails. Where to start with AI: the first case is not the most important one READ ▸ #18 02 What expires gives no warning. It just stops. None of this breaks. All of it expires. READ ▸ #19 02 The cross-cutting information already exists inside the company. What is missing is somebody saying it in business language. The area that sees all the others is the one that speaks least READ ▸ #20 02 When in doubt between two levels, treat it as the higher one. Erring upwards costs little. Before deciding who uses AI, decide what can go into it READ ▸ #21 02 A rule with no way out teaches the team to hide better. What makes it worse is not the mistake. It is hiding it. READ ▸ #22 03 An AI feature that appears in an update is a new tool. The AI you did not choose arrived in an update READ ▸ #23 TOOL · FREE TO USE Five questions that tell a supplier with a method from a supplier in the dark, in five minutes. 5 AI questions for your suppliers READ ▸ #24 01 The yardstick does not exist to make you use AI less. It exists so you stop deciding in a panic. Almost anything can go into an AI. What cannot fits in three questions. READ ▸ #25 01 Using AI at work without anyone knowing turns a company decision into your decision. You already brought AI into your company. Nobody told you what is allowed. READ ▸