The Automator

You've got agents running in more places than you've told your manager about, you think AI is great for people, and you see no reason to build AGI.

Those three facts live together comfortably.

One agent triages the inbox, one reconciles invoices, one opens the PR, one writes the postmortem for the incident it also fixed. You have a skills folder, an MCP server nobody asked for, and a quiet eval suite you're prouder of than your actual code. It's just good automation, and it's a productivity superpower.

You're the most optimistic person in the room. AlphaFold makes you misty. Narrow, excellent models curing diseases and clearing busywork are the real story, and they're already here. Today's tools already unlock huge human potential, and people are doing better work than ever.

Which is your problem with AGI: we don't need it. Some people feel near-religious awe about AI; you'd say the god is already in the tools, so why build a bigger one? Trillions in power, water and money to chase a moonshot, when we haven't finished deploying what we have? Better products beat AGI moonshots, and "AGI" is mostly a marketing term anyway. You say so in quote tweets that begin "Unpopular opinion, but."

The tension: you hand real work to autonomous software all day, then get cranky when someone suggests it might want a promotion. You don't need a god. You need an empty approvals queue by Friday.