An assistant responds to a person: it drafts, summarizes or answers, and the person decides what to do next. An agent can take actions itself — updating a record, routing a request, starting a process — within the permissions it has been given.
That difference turns AI from a productivity question into an operating question. Once software can act, the organization has to decide what it may do, who is accountable for it, and how its actions are monitored.
For executives, the useful questions are practical: which processes would benefit from an agent, which data and systems it would need, what could go wrong, and who would own it.