Companies are laying off humans and replacing them with AI agents without proper testing. Is this crazy?

I keep seeing more and more stories about entire teams being laid off and replaced by AI agents almost overnight. It feels reckless-like companies are going all-in on technology that’s still quite unpredictable in real-world conditions. Why don’t more companies take a safer approach? For example, create a parallel “AI branch” or pilot team for 3-6 months: run the AI agents alongside human employees, measure real performance, error rates, customer satisfaction, and edge cases before making permanent cuts. Is this just greed and pressure from investors to cut costs immediately? Or do companies actually have internal data showing that the risks are lower than we think? Maybe they’re seeing such massive productivity gains that they’re willing to take the gamble. I’d really love to hear from people who work (or have worked) at companies that already went through this kind of AI replacement. How did it actually go? Were there major failures, hidden costs, or surprisingly good results? What lessons would you share?