Board prep used to eat a full day or two. Now it takes 90 minutes max; no joke my budget cycles are hours, not weeks now.
People hear that and assume I found a magic tool. I did not. It is the same tool you can go buy this afternoon. The tool is not what changed. What changed is what I load into it.
That is the whole lesson, and almost everybody misses it. Including me til a few months ago.
Part one of this series was the Skeleton. Your data, built into an ontology. The nouns of your business, written down once so a machine can read them. Customer. Product. Plan. Invoice. Those are your bones.
This is part two. The Circulatory System. The blood.
Blood is context. Context is your second brain. The bones are what your business is. The blood is everything a machine needs to know to act like it actually works here.
Here is the part people get backward. They think the AI is the smart part. It is not. It's actually kind of average. I am about to tell you why.
Andrej Karpathy helped start OpenAI and built the brain behind Tesla's self-driving cars. He explains it better than anyone. A model is like a CPU. The context window is like RAM. The model is the processor, and it is fixed. It is the same processor everyone else is renting. What you load into memory is the program.
Sit with that. You and your competitor are running the exact same processor. The only thing that makes your AI different is what you put in front of it.
So what happens when you put nothing in front of it? You get the average answer. AI has read the whole internet. Ask it a question cold and it hands you the middle of everything it ever saw. Polished. Confident. Generic. It reads like it was written by a committee at a company that does not exist, because that is exactly what it is.
My boss (yes, the AI Boss has a boss) DR, also known as Danielle Rios, says AI without context is just expensive pattern matching. That is the sharpest sentence in this whole series. You can follow her here.
Think about what that does to depth. The machine knows a little about everything and nothing about you. It does not know you walked away from that kind of deal twice and both times you were right. It does not know your board only cares about three numbers. It does not know the one thing you would never do, no matter how good the math looks. It has no idea how you think, so it thinks like everybody. Everybody is not who you are competing against.
This is where guardrails come in, and I want to be plain about what a guardrail actually is.
Most people think a guardrail is a rule that stops the machine from doing something bad. That is a fence. Fences are fine. But the real guardrail is context. A machine cannot stay in bounds if nobody ever told it where the bounds are. You cannot catch a wrong answer if you never wrote down what a right answer looks like here.
Without context, there is no wrong. There is only plausible. And plausible is how smart people get talked into dumb decisions.
So you build the blood. You build a second brain.
A second brain is where you write down how you think. Not what you know. How you decide. It picks up the way you operate and sets it down inside the tool.
Here is how to build one. This is not theory. This is what sits in my tools right now. And it turned out to be too much to bury at the bottom of this issue, so I put the whole thing in a bonus issue that went out today alongside this one. It has the depth ladder, the six things that belong in the file, the folder setup, and a prompt you can paste straight into Claude so it interviews you and writes your second brain for you. Go get it.
Now the two halves come together, and this is the point of the whole issue.
There is company context and there is your context.
The ontology carries how the business works. The rules. The logic. Every tool reasoning against one shared copy, so two tools never hand you two different answers about the same customer. It is also why an agent that hides its own private copy of your rules is a trap. Vendors love that trap.
The second brain carries how you work. Your judgment, in a form a machine can read.
Both of those are context. Both have to circulate. Blood does not stop at one organ. If your judgment lives in one chat window and your business logic lives inside somebody else's tool, you do not have a circulatory system. You have two puddles of liquid.
Back to board prep. The 90 minutes is not the tool being clever. The tool knows what my board asks, which three numbers matter, what I would never put on a slide, and what broke last quarter. It has my bones and it has my blood. So it does not hand me the average board write up. It hands me mine.
That is the difference between a machine that produces and a machine that performs.
Here is your figure it out moment. Open your AI right now and ask it about a real call you are sitting on this week. Something only you would know how to answer well. Read what comes back. It will sound fine. Then ask yourself how much of it is you, and how much of it is the average of everybody.
Every line that is not you is a line you never wrote down. So what is still stuck in your head that your machine should already know?
Kathy
THE AI BOSS