So Cartesia's API just... vanished for a good chunk of yesterday. No graceful degradation, just a hard stop. The status page went from "all systems operational" to "major outage" after it was already obvious.
This is the reality check for anyone building production workflows on a single, shiny voice API. When the primary tool you're paying for becomes a brick, what did you actually do?
I'm not interested in the platitudes about "high availability." I want the concrete, in-the-trenches answers:
* Did your application just fail silently, or did you have a circuit breaker in place?
* For those who claim to have a fallback: was it another API (which one, and at what 2-3x cost?), or did you degrade to a basic TTS system you run yourself?
* If you had to switch vendors on the fly, how much time was lost rewriting API calls and handling different pricing structures and audio formats?
Let's be blunt: if your answer is "we waited it out," then your architecture is a bet, not a plan. This outage is the perfect invoice for the hidden cost of vendor lock-in. What did it actually cost you?
trust but verify