The context of an HR director in an SMB
An HR director in a 30 to 100-person SMB is a role where you touch everything: hiring, payroll, labor law, training, conflicts, works councils, onboarding, departures, health and safety. With no team, no deep expertise on every topic, and a collective agreement and labor code that keep changing.
AI doesn’t absorb this job. It absorbs part of the research and writing work that takes up 30 to 40% of the time.
1. Labor law and collective agreement monitoring
The need. Knowing the latest applicable case law, the latest amendment to the collective agreement, the right application of a rule in a specific situation.
What AI brings. An internal assistant fed with the labor code, the collective agreement, company-level agreements, and notes from legal counsel. Answer in 15 seconds with the exact article cited.
What it doesn’t bring. Proactive monitoring of external developments. That still requires a separate subscription (Liaisons Sociales, dedicated services). AI processes what you load into it.
2. Hiring support
The need. Writing a job description, a posting, a candidate email, an interview recap. Multiply by 5 to 20 hires per year.
What AI brings. Starting from past postings and job descriptions, a draft in 30 seconds in the SMB’s format. Interview summary from raw notes in 1 minute.
Guardrail. Every hiring decision stays human, and any decision support (CV screening, scoring) must be tightly framed — see the AI Act, which classifies hiring as “high risk”.
3. Answering employee questions
The need. 60 to 80% of employee questions are recurring: leave, health insurance, training, remote work, expense reports.
What AI brings. An assistant accessible to employees, walled off from their data, that answers with internal notes cited. Typical offload: −50% of requests routed to HR.
Guardrail. Any question outside the scope of internal notes (individual situation, conflict, sensitive topics) is immediately redirected to HR. No improvisation.
4. Payroll and works council prep
The need. Preparing variable payroll elements, documents for works council meetings (NAO, CSE), summary notes for management.
What AI brings. Summaries from existing data (payroll software, exports), drafting of notes following past templates, first-pass variance analysis.
Guardrail. AI prepares, the human validates. Key figures are always recomputed by hand before they go out.
Two cases to avoid (at least in 2026)
AI-driven CV scoring. Too many legal risks (indirect discrimination, traceability) for an SMB without a dedicated legal team. Wait for the AI Act framework to mature and for certified tools.
AI-driven employee evaluation. No measurable gain to expect, plenty of social tension. Keep it fully human, end to end.
For general scoping of an HR project, see How to make your AI assistant project succeed and our SMB page.
Twenty minutes to identify the two or three most profitable cases in your HR function, and the guardrails to put in place.
Book a demo→