This implementation is tune-able. One idea that comes to mind is setting up e.g. salience scores for different company-internal lingo and names of internal components and teams, enriching wiki / chats / tickets / logs this way for quicker and more accurate skimming.