Use Case HR Personnel · Munich DAX KBR · BetrVG § 80 (3) + EU AI Act

HR personnel decisions under EU AI Act Annex III - Munich DAX group works councils with § 80 (3) BetrVG expert right

Munich group works council density is Germany's highest. Statutory AI expert right consistently exercised. Group works agreement mandatory before pilot. UK Equality Act + ACAS Code parallel.

Workshop at Grindelberg All 7 use cases ←

Chapter 1 - Three Co-Determination Levers in Parallel

§ 87 + § 80 (3) BetrVG + EU AI Act Annex III - in one Decision Chain per HR personnel decision.

Munich DAX group works councils (Konzernbetriebsrat, KBR) are among Germany's most experienced negotiation partners. KBR chairs with 18-24 years of tenure. Bayern IG Metall district office Munich + ver.di Bayern + group works council exchange spread AI negotiation knowledge across sister corporates within 6 weeks. Anyone introducing an HR-AI agent without a structured Decision-Layer and audit-trail UI will be stopped in co-determination before going productive. UK group subsidiaries face the same dynamic - German entity works council holds equivalent power regardless of UK parent.

Three co-determination levers in parallel per HR-AI decision: § 87 (1) No. 6 BetrVG (German Works Constitution Act - mandatory co-determination on technical systems for behaviour/performance monitoring - AI falls automatically under this per prevailing doctrine, LAG München doctrine 4 TaBV 24/23 on working time transfers). § 80 (3) BetrVG (reformed 2021 - explicit expert right on AI, cost borne by employer). EU AI Act Annex III Point 4 (high-risk obligations for candidate selection + promotion + termination + task allocation + performance monitoring; under current law from 2 August 2026, provisionally postponed to 2 December 2027 under the Digital Omnibus of 7 May 2026 with formal adoption still pending, while the fine regime is active since 02.08.2026). UK equivalent: ICE Regulations + works council consultation + ACAS Code on AI in HR.

Plus regulatory: GDPR Art. 22 (automated individual decision with significant effect). BayLDA AI Checklist v0.9 from 24.01.2024 for GDPR compliance. DSK Orientation Guide AI and Data Protection 06.05.2024. AGG (German Equality Act) for anti-discrimination pre-check. UK parallel: UK GDPR Art. 22 + Equality Act 2010 Sec. 13-15 + ICO AI Auditing Framework 2024.

Model calculation for a DAX corporate with 90,000 employees globally (Munich Re, Allianz, Siemens central staff in Munich): ca. 400 HR personnel transactions per day. Decision-Layer split typically: 35% RULES (IG Metall Bayern tariff logic, deadline check, master-data validation), 15% AI AUTONOMOUS (document classification, skill-profile mapping, self-assessment sentiment analysis), 50% HUMAN (evaluation, promotion, selection - EU AI Act Annex III + KBR co-determination + expert consultation).

Chapter 2 - Decision Record for a Promotion Decision in a DAX Corporate

What a KBR-approved HR-AI promotion decision looks like.

Anonymised decision record for a promotion recommendation at a Munich DAX corporate (e.g. Allianz, Munich Re, Siemens). 17 candidates for grade change SG12 → SG14. KBR consultation mandatory. EU AI Act Annex III Point 4 high-risk classification. UK parallel applies for German subsidiaries of UK-headquartered groups.

PROMO-DAX-2026-05-17-Q2

Promotion proposal · 17 candidates · grade change SG12 → SG14 · quarter Q2/2026 · KBR mandatory consultation

Result 9 proposals to HR leadership · KBR consultation activated · 8 GDPR Art. 15 reasoning records generated
  1. 01 REGEL

    Eligibility check

    17 candidates in SG12. Mandatory criteria: minimum tenure 18 months, performance rating ≥ 3 in last 2 cycles, no open warning. Rule elig_promo_dax_v2.3.

    ✓ 17 eligible
  2. 02 REGEL

    Collective agreement mapping (IG Metall Bayern)

    Grade change SG12 → SG14 triggers co-determination per § 99 BetrVG (German Works Constitution Act - individual personnel measure). IG Metall Bayern tariff supplement on grade change calculated. Rule tarif_ig_metall_bayern_v1.7. UK equivalent: collective bargaining recognition + works council consultation under ICE Regs.

    ▲ KBR consultation mandatory
  3. 03 KI

    Performance classification (model <code>dax-hr-profile-v2.1</code>)

    Performance reviews 2024-2026 from SAP SuccessFactors. Sentiment analysis of self-assessment texts (DE + EN for international employees). No scoring, classification into 'strong profile' / 'profile with gaps' / 'profile unclear'. Model dax-hr-profile-v2.1.

    Confidence range 0.62 - 0.97 · threshold 0.85

    ✓ 12 strong, 3 gaps, 2 unclear
  4. 04 KI

    Skill match SG14 requirement profile

    SG14 requirement catalogue (technical skills, span of control 8-15 FTE, stakeholder management at group level) mapped against skill profiles. Model skill-match-sg14-v3.4. Top 9 with match ≥ 80%.

    Confidence 0.88 · threshold 0.85

    ✓ 9 proposed
  5. 05 REGEL

    Anti-discrimination pre-check (AGG + GDPR Art. 22)

    Statistical parity against previous-year promotion quota (gender, age, severe disability, nationality). Deviation > 15% triggers bias escalation to Compliance + KBR. Rule agg_3_v3.2 (based on AGG - German Equality Act; UK equivalent: Equality Act 2010 Sec. 13-15).

    ✓ Parity within tolerance corridor
  6. 06 MENSCH

    HR leadership selects proposals (Art. 14 EU AI Act)

    Mandatory stop per EU AI Act Art. 14. HR Director Ms M. receives the proposal list with reasoning per person + confidence score + skill-match score + bias-check status. Selects 9 proposals for KBR consultation. The agent cannot skip this step. Decision documented with timestamp and reasoning per proposal.

    ✓ 9 proposals confirmed
  7. 07 MENSCH

    Group works council consultation (BetrVG § 99) with expert right

    9 individual personnel measures submitted to KBR with the complete decision record (audit-trail UI access). KBR chair Mr K. (21 years in office) consults the IG Metall Bayern expert under § 80 (3) BetrVG. KBR has 1 week objection period. Audit-trail UI shows model version, input hash per candidate, sentiment-analysis sources, confidence score, bias-check values. Expert reviews algorithm + training-data documentation. UK equivalent: works council consultation under ICE Regs + ACAS Code on AI.

    ✓ 9 submissions · 0 objections
  8. 08 REGEL

    GDPR Art. 15 reasoning generation (8 rejections)

    8 candidates who did not make the top 9 receive, on request, structured reasoning with evaluation criteria (anonymised against others), skill-match score, bias-check values, reference to GDPR challenge right. Rule dsgvo_art15_v1.4.

    ✓ 8 reasoning records generated
  9. 09 REGEL

    EU AI Act Art. 12 logging + group works agreement audit-trail persist

    Logging obligation per EU AI Act Art. 12. Documented: training-data version, model hash (dax-hr-profile-v2.1), input hash per candidate, output, human override (HR Director + KBR), risk score, bias-check values. Plus group works agreement (Konzernbetriebsvereinbarung) audit trail with escalation thresholds + KBR consultation status + expert consultation documentation. Rule aiact_art12_bv_v1.4.

    ✓ Audit trail persisted

Chapter 3 - Workshop with KBR + IG Metall Expert

On-site workshop in Munich with works-council session in a separate room.

Engineering head office Hallerstraße 8 Hamburg. Munich workshop on-site - typically a 2-3 day block at the DAX corporate (Allianz, Munich Re, Siemens, BMW) or at Munich Urban Colab as neutral ground. Separate rooms essential: HR management + Group DPO in the main room, KBR chair + IG Metall / ver.di expert + (optional) external AI expert in a separate room with their own audit-trail UI demonstration. For UK-headquartered groups with German subsidiaries: workshop in English with remote bridge to UK HR head office.

KBR workshop pattern (5 hours): Hour 1 = Decision-Layer architecture explanation with focus on the technically enforced human-in-the-loop. Hour 2 = audit-trail UI live demo with anonymised example decisions. Hour 3 = group works agreement template walkthrough with measurable escalation thresholds + algorithm-audit-right documentation. Hour 4 = expert Q&A (often more technically deep than HR management). Hour 5 = concession negotiation (shutdown-right details, audit intervals, data access).

Group works agreement template library: we bring 4 prepared templates - (1) DAX corporate pattern (Allianz / Munich Re style), (2) Industrial pattern (Siemens / Infineon style), (3) Automotive pattern (BMW Tier-1 style), (4) UK-Parent-with-German-Subsidiary pattern (FTSE 100/250 with Munich-based DE-GmbH: parallel ICO AI Auditing Framework + UK GDPR Art. 22 audit-trail layer + ACAS Code consultation pathway for the UK head office, combined with full statutory co-determination under German Works Constitution Act + KBR Konzernbetriebsvereinbarung for the German subsidiary - both audit chains feed one Decision-Layer with two regulator-export formats). Per template: 8 mandatory clauses + optional clauses + discussed escalation thresholds. Experts receive the templates before the workshop for pre-review. The group works agreement is drafted in the workshop, not delivered afterwards.

Integration with group IT: the Decision-Layer integrates with HR-IT systems: SAP SuccessFactors (Allianz, Munich Re standard), Workday (BMW, Siemens), Personio (Tier-1 mid-market). Expert access via read-only audit-trail UI. Source code of the Decision-Layer + group works agreement configuration goes with the repository handover to the corporate - after 12-18 months the in-house HR-IT team operates the Decision-Layer without us, with KBR audit-trail access. UK parallel: source-code repository handover via IP assignment + UK GDPR Art. 22 audit trail enforceable post-handover; ICO Article 22 right of meaningful human review survives vendor transition.

Related use cases

Frequently asked questions

Why are Munich group works councils particularly experienced with AI?
Munich has Germany's densest concentration of DAX headquarters (BMW, Allianz, Munich Re, Siemens, MTU, HVB), and that produces group works council (KBR) chairs with 18-24 years of tenure. Their knowledge spreads across sister corporates within about six weeks through the IG Metall Bayern and ver.di networks. Munich corporates have also been negotiating IIoT co-determination since around 2019 (Siemens MindSphere), and the § 80 (3) BetrVG expert right on AI (reformed 2021) is consistently exercised. UK group subsidiaries face a parallel demand through the ICO AI Auditing Framework and the ACAS Code on AI in HR.
What does § 80 (3) BetrVG (expert right on AI) mean concretely?
§ 80 (3) BetrVG (German Works Constitution Act, reformed 2021): the works council has an explicit right to consult an expert on the introduction or use of AI. The employer bears the cost. In Munich DAX corporates: usually IG Metall experts (Bayern IGM district office Munich, IGM headquarters AI Working Group) or ver.di experts for insurance/services. Group works councils can also bring in external technical experts (e.g. AI scientists from TU Munich / LMU / MCML). The first works-council meeting with management about AI introduction MUST include expert consultation - otherwise the co-determination prerequisite is not satisfied. UK equivalent: ICE Regulations + works council consultation, plus ACAS Code on AI in workplace decisions; no formal cost-bearing right, but TUC + Unite have parallel expert pools for AI introductions.
What does a group works agreement template for AI-assisted HR decisions look like?
Group works agreements (Konzernbetriebsvereinbarung) for AI in HR typically contain: (1) scope (which AI systems, which HR decisions). (2) Data Protection Impact Assessment (DPIA) obligation. (3) measurable escalation thresholds (e.g. confidence < 0.85 = human mandatory). (4) audit-trail UI access for KBR + experts. (5) right to algorithm audit every 6-12 months. (6) right to data explanation per individual decision. (7) escalation path on discrimination suspicion. (8) shutdown right on breach. The Decision-Layer architecture supports all points technically - that makes the KBR negotiation feasible. UK parallel: ACAS Code on AI provides similar template, but agreement is by employer policy + works-council consultation rather than statutory co-determination.
Which EU AI Act Annex III points are relevant for HR?
Annex III Point 4: AI systems for candidate selection (filtering, evaluation, ranking), promotion and termination, task allocation, performance monitoring. Under current law the high-risk obligations apply from 2 August 2026; following the provisional Digital Omnibus agreement of 7 May 2026 they are set to be postponed to 2 December 2027 (formal adoption still pending, as of June 2026). Fine regime active since 02.08.2026 (up to EUR 35M or 7% of group turnover). Obligations: Risk Management System (Art. 9), Data Governance (Art. 10), Technical Documentation (Art. 11), Logging (Art. 12), Transparency (Art. 13), Human Oversight (Art. 14), Accuracy/Robustness (Art. 15). The Decision-Layer meets all requirements architecturally. UK parallel: UK AI White Paper 2023 + sector-specific guidance from CDEI + ICO AI Auditing Framework for HR-IT.
How is the workshop stakeholder mapping structured in Munich DAX corporates?
A Munich DAX HR-AI workshop runs 4-6 stakeholder groups in parallel: HR leadership (CHRO and HR directors), the Group DPO with Legal Counsel, the KBR chair with their IG Metall or ver.di expert (and optionally an external AI expert), plus IT Compliance, Procurement and Internal Audit. The works council cannot be skipped - otherwise the pilot will not go through. Standard venue is Munich Urban Colab as neutral ground, or the group site with a separate room for the works-council session. When a UK parent is involved, the workshop runs in English with a remote bridge to the UK HR head office.

Schedule workshop at Grindelberg

3-day discovery: Day 1 process analysis, Day 2 Decision-Layer mapping, Day 3 use-case prioritisation. Concrete deliverable.

Schedule meeting

Discovery workshop below EUR 10,000. Pilot fixed price discussed after the workshop.

← Back to overview: AI Agents Hamburg (7 use cases)