Over the past few weeks I’ve hit multiple AI/API outages — and not just one provider.
Model timeouts and HTTP 500/529 errors from Anthropic’s Claude
Repository access issues during a recent GitHub disruption
Database/API hiccups from Supabase
Even intermittent infra blips from major cloud vendors
Individually, none of these are shocking. Every system fails.
But collectively? It feels like the frequency and blast radius are increasing.
It raises some uncomfortable questions:
Are we over-optimizing for speed of development at the cost of resilience?
Are small teams unknowingly building on increasingly fragile stacks?
... continue reading