System design interview: when to mention specific technologies vs staying abstract?

There's conflicting advice on when to mention specific technologies versus staying abstract in system design interviews. On one hand, saying 'we'd use Kafka for messaging' shows familiarity and experience. On the other hand, it can derail into a debate about Kafka's internals or whether it's the absolute best choice for a given scenario, when the interviewer might just want to understand the conceptual component. How do you strike the right balance? Do you start abstract with 'a message queue' and then, if prompted, suggest a specific technology and briefly justify it? Or do you drop names early to show domain knowledge? I've found some interviewers appreciate the specifics, while others prefer a more generic architectural discussion first.