Lumo is advertised as not using conversations as training data. While I affirm this approach, individual users may want to apply some of their conversations as running context to future queries. So I propose an idea I will call the 'Conversation Vault' : A way of reserving select conversations for use as context in future discussions. Items added to the vault will be treated as part of any new conversations started until they are removed from the vault. This will allow individual users the opportunity to contextualize and personalize Lumo's behavior without submitting their content to Proton for use in Lumo's model.