01
Record
A finding Yu noticed that you reviewed and accepted.
How it works
How Yu works
Yu asks small questions, reflects back what it notices, and turns approved patterns into instructions your AI can use.
The loop
Four moves, repeated every week.
Nothing becomes an instruction until you say so. Yu offers, you decide.
01
Learn
Yu explains the idea in plain language so the next question has context.
02
Answer
You respond in your own words. Typed, spoken, short or messy is fine.
03
Mirror
Yu shows what it thinks it noticed. You approve, refine or reject it.
04
Scaffold
Accepted patterns become instructions you can carry to your AI.
Over time
A 52-week path, at your pace.
One focused topic at a time. Yu gathers repeated evidence instead of trusting a single answer, so the patterns get more precise as you go. Your assistant becomes more yours, not all at once.
Setup is quick on purpose. Your profile and a short reflection give Yu enough signal to recommend one of nine assistant models to start. From there, it customizes day by day as it learns more.
One answer can reflect a mood, a task or a bad day. Yu waits for repeated evidence and your corrections before it should get more confident.
What you leave with
Instructions your AI can actually use.
Yu does not make the model smarter. It gives your AI a clearer picture of how you work.
02
Scaffolding
The written instruction Yu builds from an accepted finding.
03
System Instructions
All your approved instructions, gathered into one set for your AI.
One example, kept short
One finding, all the way through.
01 · Noticed
You stay exploratory longer than you think.
02 · Scaffolding
While I explore, expand and test ideas. When I commit, challenge assumptions.
03 · System Instructions
Joins your other approved instructions, so your AI reads them together.
Applying your Yu automatically across providers is a future direction, not a shipped integration today. For now you copy your System Instructions into the AI you use.
For the technically curious
What should travel with you?
Yu explores the human side of portable context: what should travel, who gets to correct it, and who owns it.
This is an analogy, not a standards claim. Yu does not implement the Model Context Protocol. The comparison is about the role of a shared context layer.
