I Built 32 Stream Deck Buttons in 20 Minutes. Then I Couldn’t Explain Them.

Last updated: 2026-09-17

I generated 32 optimized Stream Deck buttons in 20 minutes.

AI produced the system. I shipped it. Then I sat there without the context to understand how it worked.

That’s a new kind of cognitive load. I call it Knowledge Debt. You get the speed. You still owe the understanding. If you never pay it back, you end up running systems you don’t understand, and they’re fragile the moment something breaks.

Finish the top 3. Then stop.

Have you noticed what you do when AI gets you through your list early?

I have. There’s a hedonic treadmill with AI productivity. You hit your top 3 outcomes faster than expected, and the natural impulse is to do more work rather than take time off. You get the win, then you fill the open space with another task.

That’s a sustainability issue. Unmanaged, it turns into overwork.

Recognize when the top 3 are done and consciously choose rest or a low-priority task. Don’t default to filling the void.

Before your next AI-assisted session, set a stop time. When you hit it, you’re done.

Document the thing you just shipped

Thirty-two optimized Stream Deck buttons. Twenty minutes. And a system I couldn’t explain.

There’s a developer debate about whether you should read AI-generated code. Same tension. You can ship something you don’t understand. Retroactive learning is how you repay the debt.

Pick one AI-generated output you used recently. Spend 15 minutes documenting how it works.

The block is no longer sit-and-execute

Traditional time-blocking assumes one task per block because human execution is linear. You sit down. You do the thing. The block is the work.

AI agents change the shape of that block. The workflow shifts to 10-80-10. You spend 10% defining the outcome. The AI executes the 80% middle. You spend 10% reviewing.

Because you’re no longer manually executing the bulk of the work, you can kick off several things in one block. A newsletter. A podcast script. A data migration. Definition and review run in parallel. The middle runs without you. Think of it as parallel processing of definition and review. You’re still responsible for the outcome. You just stopped being the person who has to execute the whole middle.

This week, identify one recurring task you can delegate to an AI agent. Give away the 80%. Keep the definition and the review.

Energy Matching beats a heroic Monday

Brooks has a Friday Afternoon Routine for tasks that are important but low-energy. He takes a regular early-Monday task and does it on Friday afternoon instead. That preserves his high-energy Monday morning for high-leverage work.

Do you know that zombie feeling of burning your best hours on admin? The task needed low energy. You spent high energy on it anyway.

Line up the energy a task requires with the energy you actually have. That’s Energy Matching.

Look at next Monday. Identify one Monday morning task that could move to a lower-energy time slot. Put it on Friday afternoon. Leave Monday for the work that needs you sharp.

Protect the week, not just the hour

I use a Meeting Day. Typically Wednesday. I cluster in-person meetings and calls into that one day so the rest of the week keeps large blocks of uninterrupted time for deep work.

How many of your deep-work blocks last week actually survived a meeting in the middle? Scattered meetings tax you twice. Once in the meeting, and again in the switching. Clustering them is how I keep the other days usable for deep work.

Not everyone can control their calendar. If you propose meetings, you can still group them on purpose.

Look at next week’s calendar. See if you can shift two meetings onto a single Meeting Day.

Brooks takes a different angle on the shape of the day. He argues that rigid, hour-by-hour time blocks are less effective than Day Theming. Instead of pinning specific tasks to specific times, he themes the morning and the afternoon. Admin & Communication in the morning. Deep Work in the afternoon. A North Star for the day, with room for interruptions and AI-driven task shifts. Less friction when the schedule breaks.

I’m more structured than Brooks is. That’s the spectrum. If hour-by-hour blocking keeps falling apart, theme your next two days instead of time-blocking them.

If you use Calendly or Outlook, use the settings to enforce Buffer Blocks automatically. Set a minimum notice period. Set mandatory gaps between meetings. Buffer time stops being a habit you have to remember and becomes a system constraint. That’s what prevents back-to-back scheduling that leads to burnout.

Go check those settings. If the gaps aren’t on, turn them on.

Let 60 days of calendar tell the truth

Want the fastest picture of where your time actually goes?

Give an AI access to your calendar for the last 60 days. Ask it to analyze the patterns. Where is the time going versus where you think it’s going? You skip the manual logging and still get data-driven insight into the leaks. This replaces the traditional time audit consultants used to charge for.

Export the last 60 days. Run the prompt.

If you only do one thing, make it the Monday morning move. Find the task that does not deserve your best energy. Park it on Friday afternoon. Show up Monday with that block still intact.