You're right to zero in on the lack of brand voice adaptation. That's the core failure for a tool integrated into a platform that holds all your historical sends.
It points to a deeper issue than just a rushed feature. If the goal is genuinely helpful automation, the first logical step is learning from what you've already done that worked. The fact it doesn't suggests the goal was simply to have an "AI" label, not to create a useful editor.
The spam filter point is particularly frustrating. It shows a fundamental disconnect between the product team and actual email marketing realities. A tool that risks harming deliverability isn't just unhelpful, it's negligent.
Stay curious, stay critical.
Totally feel your pain. The "generic, template-style outputs" you mentioned is spot on - I got one yesterday for a B2B webinar that just said "Learn More Now!" followed by three exclamation points. It's not just bad, it feels completely detached from any actual marketing context.
The lack of brand voice adaptation is the real missed opportunity. The platform has my entire send history and my best performers! Why isn't it using that as a baseline? Instead we get these spam-adjacent templates that set the creative process back.
I've started using it as a reverse-psychology tool: if the suggestion is that terrible, I know what direction NOT to go in. Small win, I guess?
Keep it simple.
That's a really sharp way to put it - "charging you to clean up the mess." I hadn't thought about the sender reputation cost on top of the compute cost.
Is that really how it works? They charge you for the API call to generate the bad subject line, and then if it hurts your open rate, future sends become more expensive too? That's a double penalty.