Raindrop (@raindrop_ai) is building the safety layer for AI agents. As agents get more capable and take on more complex work, Raindrop detects when they go wrong in production, from failed tool calls and hallucinations to problems companies didn’t even know to look for. Now they’re launching Raindrop Simulations, which brings that same approach earlier in the development process to catch failures before agents reach production. Already used by companies including Vercel, Clay, Framer, and…