Would you trust an AI more if everything stayed on your device?

One of the main reasons I built Merrin around local AI was privacy.

The more useful a personal AI becomes, the more it learns about your life: your conversations, goals, relationships, decisions, preferences, and eventually years of personal context.

I don't think that information should automatically need to live on someone else's servers.

With Merrin, we're taking the opposite approach: keep the core AI and personal context on your device wherever possible, while still letting the AI remember you and become more useful over time.

I'm curious where other people draw the line:

Would on-device processing make you more comfortable giving an AI deeper personal context, or is local processing alone not enough to earn that trust?