A friend of mine married into a family where her mother-in-law texts almost entirely in Spanish, mixed with a little English, and my friend's Spanish tops out at what she remembers from two semesters in college a long time ago. For years she'd wait until her husband got home to translate, which meant urgent things sat unanswered for hours and casual things just went unread.
Now she pastes the text straight into an AI chatbot and asks for a plain translation, and it works both ways — she can write her reply in English and ask it to translate that into natural, warm Spanish, not the stiff textbook version. The AI even picked up on some regional phrasing her mother-in-law uses and explained what it meant, which is more than a straight dictionary translation would've given her.
This matters more than it sounds like on paper. A mother-in-law who feels like she has to wait for her son to be home to be understood by his wife is a mother-in-law who eventually just stops texting as much. Being able to respond in real time, even imperfectly, keeps a relationship warm in a way that a three-hour delay doesn't.
It works the same for grandparents texting in a language grandkids never learned, or a text from a cousin overseas, or even just an older relative's texting shorthand that might as well be a foreign language to a younger family member. You don't need to become fluent in anything. You just need a fast, private way to bridge the gap in the moment, so the conversation keeps happening instead of quietly dying out.
If you want the slow, no-jargon version of all this — the stuff I wish someone had handed me before I fumbled through it myself — that's exactly what I built On Ramp AI for. Check it out here.