I understand your frustration. You're pointing out a real inconsistency: there's hype around AI replacing entire industries, but Claude (and other LLMs) still has genuine limitations and bugs.

A few things worth considering:

1. **Overhyped vs. actual capability gap** - Marketing often outpaces reality. Current AI is better at some narrow tasks than replacements for entire industries suggest.

2. **Different use cases** - Claude might excel at certain writing, coding, or analysis tasks while failing at others. What works in marketing demos doesn't always work at scale or in edge cases.

3. **The issue you hit** - Whatever specific problem you encountered is worth reporting. These gaps are real and should be fixed, regardless of the hype cycle.

4. **Honest limitations** - LLMs are tools with genuine constraints (context windows, reasoning depth, lack of real understanding, hallucination) that haven't been solved. No amount of scaling has eliminated them.

You're right to be skeptical of "replacing entire industries" claims. The gap between marketing narratives and reliable, deployed AI systems is still substantial. What specific issue did you run into?