Solution · HR & people
AI agents for HR that give your team back the people work.
HR teams answer the same questions, chase the same signatures and draft the same documents every week. AI agents can carry that load inside your HR systems, with the right permissions, while decisions about people stay with people and your works council stays on board.
Where agents pay off in HR
Most HR time goes into answering, coordinating and drafting. Agents are good at exactly that, and HR is where trust matters most.
Ask an HR team where the week goes and the answer is rarely strategy. It is the fortieth question about parental leave, an onboarding checklist across four departments and the reference letter that has waited three weeks. Necessary, repetitive, mostly text.
That is why agents fit. An agent can read your policies, look up a record with the asker's own permissions, draft in your house style and chase open onboarding tasks, then hand the result to a person for approval. Who decides does not change; the administration shrinks.
Why now: models handle long policy documents reliably, HR suites such as Personio, SAP SuccessFactors or Workday expose usable interfaces, and the rules are clearer. The EU AI Act treats AI in recruitment and employment decisions as high-risk; GDPR and German co-determination apply from day one. That tells you what to automate and what to leave to people.
- Typical starting pointsHR helpdesk, onboarding, drafting
- SystemsHRIS, ticketing, identity, e-signature, intranet
- Human checkpointsEvery decision about a person, every external document
- RegulationGDPR, EU AI Act, § 87 BetrVG
Use cases we build in HR
Each is a workflow with a system behind it and a person at the end, not a chatbot.
HR helpdesk assistant
Employees ask about leave, working time or the travel policy in the chat tool they already use. The agent answers from current policies with a citation, reads personal data only with the asker's permissions and hands anything individual to a named HR contact.
Recruiting support, without selection
From a role profile and a call with the hiring manager, the agent drafts the posting in your tone with non-discriminatory wording, then handles acknowledgements, interview slots that fit the panel's calendars and courteous rejections once a decision is made. Recruiters decide who moves forward and review anything that is not a template.
Onboarding orchestration
An accepted offer triggers a plan across IT, facilities, payroll and the team: accounts, laptop, badge, contract, first week. The agent opens tickets, chases overdue items and reports status. People approve access and sign contracts.
Internal mobility and skills mapping
Skills profiles from job history, project records and self-declared skills, with internal openings or learning paths suggested to employees who opt in. The agent suggests and never ranks people for a role; managers and HR decide, and employees see what is stored about them.
Surveys, documents and HR reporting
Open-text survey answers become themes with anonymised quotes. Reference letters and Arbeitszeugnisse are drafted in the correct form and always signed off by HR. Monthly headcount, absence and attrition summaries come from your HRIS, every number traceable.
Worked example
How an onboarding agent runs a new hire's first month
A typical scenario: a mid-sized manufacturer hires eight to twelve people a month, and onboarding lives in a spreadsheet across four departments. With an agent it looks like this.
- Trigger. The signed contract in the HR system starts the workflow. The agent reads role, location, start date, manager and cost centre and picks the matching template.
- Plan. It creates the task list: accounts and laptop, building access, payroll data, safety briefing, first-week meetings, each with an owner and a due date.
- Requests. Tickets open in the IT service desk and the facilities system, pre-filled with the role profile. Access is requested, not granted: an IT administrator approves it. First human checkpoint.
- Documents. The agent drafts the welcome message, first-week schedule and equipment handover form. The manager edits and confirms the schedule. Second checkpoint.
- Follow-up. Each morning the agent checks open tasks, nudges owners of overdue items and escalates to HR after a defined delay. It never chases the new hire.
- Day 30 and the log. The agent collects a short check-in from manager and employee for HR. Every action, request and approval has been recorded, so HR can show who approved which access and when.
Building this takes weeks, not months, if your HRIS and ticketing systems have usable interfaces. The same pattern covers offboarding, where disciplined revocation of access matters more than the time saved. See agentic workflows and knowledge assistants for how we build this.
Guardrails and risks in HR
Recruitment and selection are high-risk under the EU AI Act
AI that filters, ranks or evaluates candidates or employees is high-risk, with documentation, oversight and transparency duties. Our recruiting agents draft, schedule and inform; automated selection or ranking is off the table. Obligations phase in and timelines may shift, so check the current status with counsel.
Co-determination first, not last
Systems that could monitor performance or behaviour trigger the works council's rights under § 87 BetrVG. We involve the council early and document what the agent can and cannot see. An agent introduced over the council's head gets switched off.
Personal data on a need-to-know basis
The helpdesk answers with the permissions of the person asking: managers see their team, employees their own record, nobody a salary they are not entitled to. Special categories such as health data stay out of reach unless a documented purpose requires them.
Transparency towards employees
People know when they are talking to an agent, what it uses and how to reach a human. The AI Act's transparency duties and its AI-literacy requirement (Article 4) are part of the rollout, with short training for HR and managers.
Confident wrong answers about entitlements
A wrong answer on notice periods or leave does real damage. We ground answers in versioned policies with citations, test against a question bank before launch and monitor low-confidence answers in the first months, longer with managed AI operations.
How we start
Assessment
We map HR processes, ticket volumes and question types, review your systems, and rank two or three use cases on value, data readiness and regulatory risk. See our AI readiness assessment.
Design and alignment
Workflow design with your HR team, the works council and your data protection officer at the table: permissions, checkpoints, escalation rules and the metrics we measure against a baseline.
Pilot
A production-grade pilot in six to eight weeks, integrated with your systems, tested on real questions and real onboardings and measured against the baseline, with an HR owner who can adjust policies and prompts without us.
Scale
Further use cases follow the same pattern with the governance in place, plus training for HR and managers and, if you wish, managed operations as policies change.
Frequently asked questions
Is an HR helpdesk agent compatible with GDPR?
It can be, if designed that way. General policy questions are answered from documents without personal data; individual lookups use the asker's permissions, a documented legal basis and logging, with an impact assessment where required. We prefer EU hosting or on-premise models for HR data. Not legal advice; your DPO has the last word.
Do we need to involve the works council?
In Germany, almost always. Any system that could technically monitor performance or behaviour is subject to co-determination under § 87 BetrVG, and most HR agents touch that line. Early involvement tends to make projects faster: the council's questions are the ones a good design answers anyway.
Can the agent screen or rank applicants for us?
We advise against it and do not build automated selection: it is high-risk under the EU AI Act, and candidates have rights regarding automated decisions under GDPR. We build structured application summaries, scheduled interviews and courteous communication instead, so recruiters spend their time on judgement.
What happens to the jobs in our HR team?
Honestly: the administrative share of HR work shrinks and the share that needs judgement, conversation and presence grows. Most companies are understaffed for the second kind, so the effect is a shift, not a cut. Where roles change, we help HR plan the transition.
Related pages
Enterprise knowledge assistants (RAG)
Assistants that answer from your SharePoint, Confluence, DMS, ERP and ticket history: permission-aware, with citations, evaluated, hosted in the EU.
Agentic workflow automation
Multi-step business processes automated end to end by LLM-powered agents across ERP, CRM, ticketing and email, with human approval steps built in.
AI governance, EU AI Act & GDPR
An AI register, risk classification under the EU AI Act, GDPR-aligned processes and a usage policy your teams will actually follow, built together with your lawyers and your data protection officer.
The EU AI Act for companies
Who the regulation applies to, which risk tier your use cases fall into, what has applied since 2025, and a seven-step plan to get compliant without drama.
Let's find the first workflow worth automating.
A 30-minute intro call, no slides and no obligation. We listen, ask about your processes, and tell you honestly where AI agents would pay off and where they would not.