Not everything should be automated. The skill isn't using AI—it's knowing when to use it.
There's a temptation, when you first discover what AI can do, to automate everything. Draft every email. Summarize every document. Generate every first draft. It feels like unlocking a superpower.
Then you realize something uncomfortable: some of those tasks were actually valuable. The thinking that happened while drafting that email? That was where the insight came from. The friction was the feature.
Good managers know that delegation isn't about offloading work—it's about matching tasks to the right capabilities. The same principle applies to AI.
Tasks that are repetitive, rule-based, or require synthesis across large amounts of information.
Tasks where the process itself creates value—where thinking through the problem is the point.
Tasks where you need a starting point, but your judgment shapes the final output.
The third category is where most knowledge workers should spend their time.
"The best way to have a good idea is to have a lot of ideas."
— Linus Pauling
AI gives you more ideas to work with. The selection is still yours.
// The collaboration is the competitive advantage
What percentage of your time goes to high-value thinking vs. low-value processing?
Healthy: >60% on judgment-heavy work
Broken: <30% on strategic thinking
Are you still struggling with hard problems, or has AI made everything feel easy?
Healthy: Regular "productive struggle" moments
Broken: Can't remember the last hard problem you solved alone
Understanding what to delegate versus what to keep for yourself.
The most effective AI usage involves a back-and-forth. You start with your own thinking, use AI to expand or challenge it, then synthesize the result with your judgment.
You're preparing a proposal. Instead of asking AI to write it, you write the key points yourself. Then you could ask AI to identify what's missing, suggest counterarguments, or find supporting data.
The final proposal would be yours, but it would be better for having used AI as a thinking partner.
Before delegating a task to AI, ask yourself: "If I never did this task again, would I lose something important?"
If the answer is no—delegate it. If the answer is yes—keep it, or collaborate. The friction might be the point.
The goal isn't efficiency at all costs. The goal is spending your time on work that matters.
Let's look at how you spend your time and identify where AI could genuinely help—and where it shouldn't.