Can We Trust AI as Customer Support?

I’m curious how founders, engineers, and support folks here think about this.

A lot of customer service is moving toward AI agents—handling FAQs, troubleshooting flows, and sometimes even resolving tickets end-to-end.

But can we fully trust AI to act as front-line customer support?

If yes, what convinced you? If no, what are the biggest blockers—accuracy, edge cases, empathy, accountability?

Would love to hear real experiences from teams that have tried replacing (or augmenting) human support.