Claude is down? - what's your contingency plan?
After all Claude sessions stopped for 10-15 minutes around 11:15 CET, the thread discusses contingency plans so work does not halt when a provider goes down. The practical setup: the same task can run on two providers, session memory lives outside either of them so switching does not mean starting over, and the fallback is kept warm with a monthly drill where a real task actually runs on the backup.