Most of what people mean by "automating a job" is automating the repetitive, rule-based parts of it: pulling the same report every week, sending a follow-up, checking a dashboard for problems, drafting a first version of a page. An agentic system can run that whole loop on its own: read the situation, decide what to do, act through the tools already in use, check the result, and retry or flag a person when something is off. That removes hours of grinding work. It does not remove a person by itself.
What a system still cannot do well is judgment calls with real stakes, relationships tied to a specific person, or situations that do not match a known pattern. We map the manual work before building anything and say plainly which parts a system can own and which still need a human. Most roles turn out to be a mix: the repeatable half moves to the system, the half that needs a decision stays with a person, who now has more of the day for it.
Whether that adds up to a layoff is a business decision, not something the technology decides for you. A solo owner or small team usually uses automation to avoid hiring for grinding work, not to remove someone already on payroll. A role built almost entirely out of repeatable tasks is a different case: automating it does remove the need for that specific role, and it would be dishonest to say otherwise. What happens to the freed time, whether that means growth work or a smaller team, is worth deciding on purpose.