[Fruited AI] Implement a 'Persistent Context System'

The model already knows how to write an email or summarize a document. You don't need to teach it that. What you actually need to give it is context: who you are right now specifically, what you're working on this week, what decisions you've already made that aren't up for reconsideration, what your communication style is. That's the gap between a generic AI response and something that actually sounds like it comes from someone who understands your situation. The "decisions already made" framing is the most underrated part. Without it the assistant tries to be helpful by reconsidering things that aren't up for reconsideration, which is a massive time sink. And specificity beats formality every single time: "this person interprets silence as agreement so I want to be explicit that this is not a yes" is infinitely more useful than "write a professional response." The model doesn't need coaching on tone, it needs actual information about the situation. The logical endpoint is that prompting a personal assistant well is really about maintaining a persistent context layer, not crafting individual prompts. The better the assistant's ongoing model of who you are, the less work you do per interaction. Most tools still aren't designed this way in 2026 which feels like the obvious next frontier. Anyone building their own persistent context system or using something that actually handles this?