I started writing cloud notes because too many small details were disappearing between tickets.
A redirect rule for a renamed service was missing from the diagram. A temporary NAT instance still had an owner field and a passed review date months after it should have been removed. The same redirect question appeared in more than one ticket because the answer had never been written where the next person would look.
None of those problems were dramatic on their own. They were the kind of small cloud details that become expensive when they are rediscovered under pressure. Once the expiry date for the NAT instance was visible in the checklist, removal took one afternoon. The hard part was not the change. The hard part was noticing it had become invisible.
That is what the notes are for. Not polished essays. Not a perfect knowledge base. Just enough written context to preserve the thing that would otherwise be rebuilt from memory.
A good note saves the next person from asking the same question with less time and more pressure.