Been thinking about this as my team deploys more AI agents across different functions.
We've got agents handling customer support triage, code reviews, content drafting, data analysis. Started with one, now we're at maybe a dozen running in various capacities. And I'm realizing... nobody's actually managing them as a coherent workforce.
- Who decides which agents we use vs. build vs. skip?
- Who tracks if they're actually performing well or just generating confident-sounding garbage?
- Who notices when one agent's outputs conflict with another's?
- Who owns the security/permissions picture across all of them?
Right now it's ad hoc. The dev team manages the coding agents. Marketing manages theirs. Everyone configures things differently. Nobody has the full picture.
It feels like early-stage companies before HR existed—just a bunch of people doing their own thing until you hit a scale where the chaos becomes unsustainable.
The weird thing is AI agents aren't quite "tools" (they're too autonomous) but they're not quite "employees" either (no motivation, different failure modes). It's a new category.
Anyone else thinking about this? How are you managing your AI agent footprint as it grows? Is this eventually an IT function, an HR function, or something new entirely?
[link] [comments]