What does the myth say?
That an AI employee is a person-shaped substitute: hire one, and one person is no longer needed. The framing comes from the name, from vendor decks that price against salary, and from the fear that any tool called an employee is designed to make an employee redundant. It is the objection most leaders hear first from their teams and the one that most often stalls a pilot before it starts.
It is worth examining honestly rather than dismissing, because there is something true in it, and pretending otherwise is how trust with a team is lost.
Where does the belief come from?
From three places. The vocabulary: employee, hire, FTE, all borrowed from headcount. The pricing conversation: vendors, including this one, compare cost per outcome against what the outcome cost before, and the before was usually a person's time. And the experience of previous automation, where a tool arrived and a job was later cut. People are not wrong to notice the pattern; they are wrong about which part of the job the tool takes.
The vocabulary is used here deliberately, because the management model, a direct report that reports and is reviewed, is the accurate one. The consequence, replacement, does not follow from it.
What is true in it?
Two things. First, a Digital FTE does absorb hours that were a person's: the reconciliation, the screening, the first-line replies, the CRM hygiene, the drafting. Those hours do not come back as the same work. Second, where a company was about to hire for repetitive work, a Digital FTE can absorb that hire, and the person who would have been recruited is not. That is a real effect on a real hiring plan, and it should be said plainly rather than hidden inside redeployment language.
What is not true is that the person who held the role is no longer needed, because the role was never only the repetitive part.
What does the mechanism actually do with a role?
It splits it. Every task class the Digital FTE on ftes.ai handles starts at Level 0 on the autonomy ladder: the FTE proposes with evidence, and a person approves, sends back with a reason, or escalates. Routine classes earn autonomy from the approval record, one class at a time, within bounds; decisions about people, prices, money, and anything on the always-escalate list never do, at any level, by design rather than by setting. The person who held the role becomes the approver and the manager, and the judgement work that was squeezed into evenings becomes the day.
The morning brief, the approval queue, and the send-back reasons are the person's new tools. The person's name is on every decision in the audit trail.
- FTE: proposes repetitive, evidence-based work with the evidence attached
- Person: approves, sends back, escalates, decides, manages
- Never automated: decisions about people, prices, money, escalations
What usually happens to the team?
The team is redeployed toward the judgement work: the finance person moves from matching to exceptions and analysis, the support agent from password resets to complaints and account problems, the recruiter from screening to interviewing and closing, the marketer from drafting to positioning. Some people like this more than the old job; some need help with the change, and that conversation is the manager's and goes better early. The headcount line and the AI line buy different work, and the honest framing to a board is allocation rather than replacement.
Where a hire was planned for repetitive work, it may not be made. That is the honest exception, and it should be said to the team as such.
How can you check for yourself?
Sort one role's work for a fortnight. Every task goes in one of three columns: repetitive with a checkable right answer; judgement, relationship, or a decision about a person; or the approval step between them. Count the hours. The first column is what a Digital FTE would take; the second is what stays; the third is what the person does instead of the first. If the second column is small, the role was mostly repetitive work and the honest conversation is about what that person does next. If it is large, which it usually is, the myth is disproved by your own sheet.
Run the same sheet after a month with the FTE and compare the columns. The baseline-time template in the library is built for exactly this.