Where does the line fall in HR?
In HR, the line between what a Digital FTE can own and what it cannot falls between administration and judgement about people. Screening against a scorecard, scheduling interviews, running onboarding checklists, and answering handbook questions are administration. Deciding who to reject, hire, promote, or discipline is judgement about a person.
A Digital FTE on ftes.ai is built with that line fixed. Actions that change a person's outcome are permanently gated and never earn autonomy, because the cost is asymmetric and accountability must sit with a named human.
What can it own in recruiting?
In recruiting, the Recruiter Digital FTE screens applications against the scorecard you set, produces a ranked shortlist with the evidence behind each score, drafts candidate communication for approval, and schedules interviews into connected calendars. Interview scheduling can earn autonomy because it changes no outcome.
Rejections and offers are drafted for a person to send, if at all, and the audit trail records who decided. Candidates who need a human conversation before ranking are flagged rather than scored.
What can it own in onboarding and policy?
In onboarding, a Digital FTE can run the logistics: account requests raised, equipment and access checklists tracked, first-week schedules drafted, and reminders sent for approval or, once earned, automatically. The new hire's manager sees what is done and what is waiting.
For policy questions, it answers from your handbook and policy documents with a citation to the section it used, and escalates anything the documents do not cover or anything sensitive, such as a grievance, a health matter, or a question about a specific person's situation.
- Screening: scorecard-based, evidence per score, human decides
- Scheduling: into connected calendars, can earn autonomy
- Onboarding: checklists and reminders, manager sees progress
- Policy answers: cited from the handbook, sensitive topics escalated
What is permanently human?
Rejections, offers, compensation decisions, performance ratings, disciplinary action, grievance handling, and anything involving a person's health, family, or protected characteristics are permanently human. The Digital FTE may prepare information for these, but it does not decide and it does not communicate the decision.
This is fixed in the design rather than a setting. An HR leader evaluating any AI for the function should ask which actions can never be automated and expect a clear list.
Where should an HR team start?
Start with interview scheduling and first-line policy questions. Both are high-volume, low-risk, and easy to measure: time from shortlist to scheduled interview, and share of policy questions answered at first line with a citation. Neither touches a decision about a person.
Add screening once the scorecard is written and the team has reviewed a few ranked shortlists with their evidence. The how-to-hire guide for HR on ftes.ai covers scorecard design and the audit approach for fairness.
How is the HR data kept safe while the FTE works?
HR data is kept safe by scoping what the Digital FTE can reach and by never letting it decide about a person. The Recruiter FTE connects to the applicant tracking system per property with the narrowest permissions that allow screening and scheduling, and the policy-question FTE connects to the handbook and policy documents rather than to personnel records. Connections are encrypted before storage, never appear in logs, and can be disconnected from the workspace at any time, with the change logged.
Anything about a specific person's situation escalates to a named HR team member rather than being answered, so sensitive personal data does not become the basis of an automated reply. The audit trail records every screening score with its evidence, every communication approved with a name, and every escalation with who received it, which is the record a fair-process or data-protection review asks for. Before connecting any system holding personal data, the data protection lead should confirm the processing basis and the vendor's data handling terms in writing; the GDPR guide on ftes.ai lists the questions.
Record the baseline before the first proposal: time from shortlist to scheduled interview, share of policy questions answered at first line, and the hours the HR team spends on each. Two cycles later the same measures, read from the Targets tab with evidence, are the comparison, and they show which of the next functions to add and which to keep exactly as they are.