I Have Unlimited Execution Now. My Attention Ran Out First.

After a week of working with agents the way I work now, I don't want a strategy game in the evening. I want something low stakes that I can just watch.

That's new for me, and it took me a while to see where it comes from.

For the past four months I ran agents, subagents and workflows across almost everything: projects, small apps, improvements to my own agent setup. Some days I had 50+ threads open at once. Output-wise it was the best stretch I've ever had.

Then the bill started arriving at the end of each week.

A hundred tiny decisions

It wasn't physical. Every thread comes back at some point: approve this, test that, go left or right. A hundred teeny tiny decisions a day, with a context switch between each one. It felt like being in meetings all week.

A few things I learned from it:

• The ceiling moved. Tokens, model intelligence and compute used to cap what I could get done. They all dropped below it. My attention, direction and time are the limit now.
• Where you sit in the loop matters. Mine is mostly the end: my agents pull context from what I'm already doing, the middle is theirs, and I come in to test, approve and check quality. One check is nothing. Fifty a day is where the tiredness comes from.
• It has a name. BCG surveyed 1,488 full-time US workers and called it "AI brain fry". People doing heavy oversight of AI reported 12% more mental fatigue.
• Most people haven't hit it yet. With one chat and one thing at a time, decisions arrive one by one. With many threads across many projects, they arrive all at once.

What I'm doing about it

The full post goes further: what my role is when execution is basically sorted, how work changed shape from pre-computer offices to multitasking to 50+ agent threads, what happens when even the decisions go to AI, and the thing I'm building to keep all those decisions in one place.

Read the full post on Digital Thoughts: I Have Unlimited Execution Now. My Attention Ran Out First.

If you run many agents in parallel, I'm curious: what eats most of your decisions, approvals, direction calls, or checking quality?