I’m working on something to make non-dumb AI Agents easier to create and deploy. Now, we're researching how teams integrate AI agents into daily workflows. It seems like AI buzz is everywhere, but it's still hard to adopt it and then make it work for real processes.

If your company (1-50 ppl or a series A/B startup) tried AI agents for any purpose, or considered it but dropped the idea, what's your experience?

What worked, what didn’t? Biggest friction (setup, prompting, cost, trust)? Did you keep it in production or roll back — why?

Trying to map where agents add value vs. where they stall now.