The assistant fantasy is mostly about consistency

What keeps feeling missing to me isn’t just better answers, it’s a sense that the thing will stay itself tomorrow. The sci-fi versions were basically characters with a stable voice and memory; real AI tools still feel more like a feature flag that gets swapped out while you’re using it.

Maybe that’s why people forgive a lot when the assistant is predictable, even if it’s not that smart. Has anyone else found they care more about consistency than raw capability?

What you’re describing maps pretty cleanly to “API stability, ” just for personality: if the assistant’s tone/rules/memory change under you, you can’t build habits or workflows around it.