Prompting
Custom instructions
User-supplied standing preferences or instructions applied across supported interactions.
Example
A user asks an assistant to prefer concise explanations.
Why people use it
It saves people from repeating the same preferences in every conversation.
What you'll hear
“Keep explanations short unless I ask for more detail.”
What this means for you
Avoid placing secrets in persistent preferences and check that important instructions were followed.
Can you control it?
Sometimes
Sometimes. Your choices depend on the tool and your access. The settings available to an everyday user may differ from those available to the people running it.
Common questions
- Do custom instructions override every other rule?
- No. Higher-priority system rules and product behavior still apply.
- Can old instructions become unhelpful?
- Yes. A preference suitable for one kind of work may interfere with another.
- Does adding a preference teach the basic AI permanently?
- Not by itself. The service supplies the instruction when it uses that preference.