I try to make a clear distinction between:

1) Questions for which you can recall the answer off the top of the head, or by sharing an _existing_ note, document, link, etc.

2) "Questions" which cannot be immediately answered from memory / existing records, and are in fact a request to do _new work_, be it research, analysis, writing some document, etc. etc.

Hopefully, you are sufficiently well organized, keep notes, and anticipate obvious inquiries such that that many questions are of type #1. Think of those as L1 or L2 cache hits.

Then, for the remainder that are #2, you can say "sorry, I don't have anything to hand, haven't thought about that, but I could look into it and get back to you with something in X amount of time, if that's useful. should we create a ticket, and prioritize this alongside other priorities?".

The thing that will inspire confidence is not saying this all the time, but only for the non-#1 things, and that many things are #1 and get an immediate response.

It's also powerful to build these constructs into questions of other people. You could ask "Hey, Bob - I'm wondering, do you know, off the top of your head, where the code that does X is?".

Ideally, Bob can then say, "oh, yes, I was just looking at that yesterday, it's here: http://....", or alternatively "Hmm. Actually, no, sorry.".

An annoying non-answer would be "No, but it's probably in place X, because Y, or maybe it's Z, or it could be found using git history or blah blah blah..." - that's not helpful, because if it needs to _searched for_, you can do that just as well as Bob, the question was whether he had it _to hand_, not whether he could make some guesses about where it _could_ be...