Keep your own thinking: don't delegate metacognition

The most useful AI habit is not asking for an answer. It is using the answer as material for better thinking.

AI can draft an email, compare options, summarise a meeting or generate ten ways to structure a project. It cannot know which customer relationship matters most, what level of risk you are willing to accept, or what kind of business you want to build. Those are judgments, not missing information.

Separate preparation from judgment

This distinction is practical:

Let AI help with	Keep ownership of
Gathering facts and drafting options	Defining the actual problem
Finding inconsistencies and counterarguments	Setting decision criteria
Turning notes into a clear brief	Choosing the trade-off you accept
Preparing a first version of a plan	Committing resources or communicating a final decision

The boundary is especially important when a decision is expensive, difficult to reverse or affects another person. AI can make those decisions look neat long before they are wise.

A five-minute decision practice

Before you ask an assistant for advice, write three lines in your own words:

1. What decision am I actually making?
2. What must be true for this to be a good decision?
3. What would make me regret it in six months?

Then ask the tool to challenge your assumptions, identify missing information or propose alternatives. When it returns, write a short decision note: what you chose, why, and what evidence would change your mind. This is enough to preserve the thinking trail without turning every decision into bureaucracy.

Use AI as a sparring partner, not an authority

Good prompts invite disagreement: “What is the strongest case against this plan?”, “Which assumption am I treating as fact?” or “What would a cautious operator check before acting?” The goal is not to make the model sound certain. It is to make your own reasoning more visible.

For repetitive work, you can delegate more. For strategy, customer promises, hiring, pricing and priorities, keep a human owner who can explain the rationale and live with the outcome.

Build a record you can revisit

A second brain is useful here because it holds the context of your decisions over time. A short note can reveal whether you repeatedly avoid the same issue, whether an assumption was wrong, or whether a past choice produced the result you expected. That is learning an AI chat history cannot do for you by itself.

AI should give you more time and better inputs. Use that gain to think more clearly, not to skip the part of the work that only you can do.

Next steps

Learn how to build a second brain in Obsidian for AI, then use the D.A.D. + V framework to decide what is worth removing, automating, delegating or validating.