Absence is not a calendar problem. It is a decision problem.
30+ leave types. Multi-country collective agreements. 78-87% automatable.
The AI Agent classifies the leave type, validates entitlements against collective agreements, company governance frameworks, and law, detects team conflicts, and monitors return-to-work thresholds - more reliably than any HR administrator. The calculations themselves - remaining entitlements, sick pay deadlines, part-time pro-rating - run through deterministic rule engines. The human stays in the process where employee representation law, employment law, or health data protection demand it.
Leave types (statutory, collective, company)
Micro-decisions per transaction
Zero-touch rate (simulation)
Leave types: statutory leave law, sick pay regulations, maternity/parental leave, carer's leave, severe disability additional leave, collective agreements + company governance frameworks.
The problem
Why absence systems fail in complex organisations
SAP SuccessFactors Time Off records absences. Workday Absence calculates balances. Every HRIS can model approval workflows. But in organisations with multiple collective agreements, multi-country operations, part-time models, and return-to-work obligations, recording is not the problem. The decision upstream is: which entitlement applies? Which policy takes precedence? Must a return-to-work process be initiated? (US: No federal mandatory leave law. FMLA provides 12 weeks unpaid. State laws vary significantly.)
30+ leave types
Annual leave, special leave, educational leave, maternity leave, parental leave, carer's leave, child sick days, severe disability additional leave, collective-agreement time-off options, flexitime days, company shutdowns, sabbaticals. Each type has its own legal basis, its own deadlines, its own calculation rules. No administrator knows all rules across all collective agreement zones and jurisdictions.
Every sick note triggers 5 follow-up processes
Process the digital sick note, check sick pay continuation deadlines (typically 6 weeks per case), monitor return-to-work thresholds (42 sick days rolling), notify shift planning, write back to payroll. Manually: media breaks across five systems. Forgotten return-to-work triggers. Incorrect sick pay limits.
Zero decision transparency
When employee representatives ask: Why was Employee X's leave rejected? Which rule applied? Who decided? The evidence is missing. Absence systems document outcomes - approved or rejected requests. Not the decisions that led to them.
The Decision Layer
Every absence transaction. Broken into decision steps.
The Leave & Absence Decision Layer breaks every absence transaction into individual decision steps. For each step, it defines who decides: the AI Agent classifies leave types, detects conflicts, and monitors thresholds - more reliably and faster than any administrator. The rule engine calculates entitlements, deadlines, and conversions - deterministically and reproducibly. The human stays in the process where employee representation law, employment law, or health data protection demand a human decision.
| Step | Decision | Decision maker | Rationale |
|---|---|---|---|
| 1 | Classify leave type | AI Agent | Agent identifies from request or free text: annual leave, special leave, educational leave, flexitime, child sick day? Assigns correct legal basis |
| 2 | Calculate entitlement (hierarchy) | Rule engine | Statutory leave law + collective agreement + company governance framework + employment contract = total entitlement. Apply most-favourable principle |
| 3 | Determine remaining entitlement | Rule engine | Entitlement - taken - planned = available. On part-time change: prorated recalculation per case law |
| 4 | Check team conflict and minimum staffing | AI Agent | Agent checks team calendar, company shutdowns, minimum staffing. Detects overlaps and calculates available capacity |
| 5 | Generate approval recommendation | AI Agent + Rule engine | All rules met: "Approval recommended" to line manager. Forwarded with context and recommendation |
| 6 | Escalate leave conflict | Human | Two employees, same week, minimum staffing breached. Social considerations (school-age children, single parents) require human judgement - employee representation law demands it |
| 7 | Calculate sick pay continuation deadline | Rule engine | Statutory sick pay period per case. On recurrence: waiting period and prior-employment rules apply |
| 8 | Check return-to-work threshold | Rule engine | 42+ sick days in 12 months (rolling). Automatic trigger to HR and employee representatives. Health data separated |
| 9 | Process digital sick note and write back | Rule engine | Digital sick note from insurer: book absence, notify payroll and shift planning, update calendar |
| 10 | Generate audit entry | Rule engine | Decision, rule basis, decision maker, timestamp, input hash - append-only, SHA-256 signed |
5 to 15 steps per absence transaction. The AI Agent handles every one of them better and faster than a human administrator. Yet the human stays at defined points in the process - not because they do it better, but because employee representation law, employment law, or health data protection demand it. At 10,000 employees, that means 50,000+ documented micro-decisions per month.
AI classifies. Rule engine calculates.
The AI Agent identifies: which leave type? Which legal basis? Is there a team conflict? Has the employee reached the return-to-work threshold? This classification is more reliable than any human administrator. The calculation of entitlements, deadlines, and conversions then runs through versioned Decision Tables - deterministic, reproducible, auditable.
Return-to-work data architecturally separated
Health data is a special category (GDPR Article 9). No mixing of leave data and diagnoses. The return-to-work module has a dedicated database, dedicated access circle, dedicated retention periods. The absence agent sees only: return-to-work obligation triggered yes/no. No diagnoses, no sick note reasons.
AI Agent, rule engine, or human
The AI Agent classifies facts and detects patterns - more reliably than any administrator. The rule engine calculates entitlements and deadlines. The human stays in the process where employee representation law, employment law, or health data protection demand a human decision.
Leave forfeiture notification automated
Case law requires individual, timely warning when leave is about to expire. Without proof of notification, leave never expires. The Decision Layer generates the notifications automatically, documents delivery, and escalates when proof is missing. (UK: UK statutory leave carries over automatically in many circumstances.)
Governance
Not documented after the fact. Created in the process.
When employee representatives ask "Why was the leave request rejected?" - "minimum staffing" is not enough. Which staffing rule? Which period? Who was already approved? The Leave Decision Layer generates a decision record for every absence transaction that answers exactly these questions.
Every micro-decision generates an audit entry
Append-only. Nothing is overwritten, nothing deleted. Corrections create reversal and adjustment entries. SHA-256 signed, exportable at any time.
Employee-representation compliant
Employee representation bodies hold co-determination rights over leave policies under applicable employment law. Rule engine transparent, rejection reasons documented, anomaly detection controllable via feature flag, reports pseudonymised.
Co-determination →GDPR Article 9: health data separated
Return-to-work module architecturally isolated. Dedicated database, dedicated access circle. No diagnoses in the absence module. Return-to-work files separate from personnel file. Retention periods configurable.
Data residency →Leave forfeiture notification automated
Case law establishes that leave does not expire without individual, timely notification. The Decision Layer generates notifications automatically, documents delivery, and escalates when proof is missing.
Integration
Connects to your existing system landscape
Your systems stay. The manual decision work upstream disappears. The Leave Decision Layer sits between your source systems and your HRIS - it makes the decisions that administrators make today.
Data sources
- →Employee master data (SAP HCM, Workday)
- →Time management (Kronos, ATOSS)
- →Digital sick notes (insurer feeds)
- →Shift planning (ATOSS, SAP)
- →Collective agreement database
Leave Decision Layer
- Leave type classification
- Entitlement calculation (CBA + company + law)
- Team conflict and staffing checks
- Sick pay continuation calculation
- Return-to-work threshold monitoring
- Leave forfeiture notification
- Digital sick note processing + write-back
Target systems
- →SAP SuccessFactors Time Off
- →Workday Absence Management
- →Payroll (ADP, SAP HCM)
- →Calendar (Microsoft 365, Google)
- →DMS / Digital personnel file
Simulated for enterprise volumes
Four industries. Calculated.
We configured the Leave & Absence Decision Layer for four industries with realistic collective agreements, absence rules, and workforce structures. Each card shows the simulation parameters and the result.
Chemicals
- Collective-agreement time-off options
- Shift-supplement leave 2-5 days (continuous operations)
- Hazardous materials return-to-work obligations
- Site-level collective agreements
Aviation
- Multiple bargaining units (cockpit, cabin, ground)
- EASA FTL: mandatory rest periods, duty-period limits
- Medical Grounding (loss of fitness certificate)
- Crew rotation across bases
Financial Services
- Regulatory mandatory leave (four-eyes principle, fraud detection)
- Compliance-mandated leave blocks for traders
- Banking holidays in addition to public holidays
- Phased retirement models
Retail
- 6-day working week: statutory minimum 24 working days
- 64% part-time ratio, variable-hours contracts with full entitlement
- Seasonal blackout periods (Christmas, Easter, peaks)
- Regional collective agreements
Implementation
From pilot to production.
Technical architecture
The Leave & Absence Decision Layer runs entirely within your infrastructure: your data centre, your network, your sovereignty. No SaaS dependency, no data exfiltration, no external telemetry tracking. Health data and return-to-work files never leave your network. Containerised, multi-tenant capable, deployment-ready for your private cloud.
Implementation
The Decision Layer is not installed - it is configured: your collective agreements, your company governance frameworks, your absence types. Typical pilot projects launch within 3 months with one site and the most frequent leave types. Extensions to additional sites, collective agreements, and special cases run in parallel with pilot operations.
Economic levers
Eliminate return-to-work deadline failures
Without a properly conducted return-to-work process, a long-term sickness dismissal is almost always unenforceable. The Decision Layer detects the threshold automatically, triggers the process, and documents the invitation. No forgotten obligations.
Digital sick note processing in real time
The Decision Layer processes digital sick notes automatically: book absence, calculate sick pay continuation deadline, notify payroll, update shift planning. No manual data entry across five systems.
78-87% zero-touch. Human only where legally required.
Standard leave requests, remaining entitlement calculations, special leave, digital sick note processing, child sick day quotas: all automatable. Human intervention remains for leave conflicts, return-to-work meetings, and educational leave content review.
Eliminate leave forfeiture risk
Case law establishes that without individual, timely proof of notification, leave never expires. Limitation periods only begin once notification obligations are met. The Decision Layer automates notifications and documents delivery.
Security
Enterprise security. From day one.
Absence data contains health information, return-to-work files, and sensitive personnel decisions. The Leave Decision Layer is designed for regulated environments where data protection, audit readiness, and traceability are not optional extras.
100% customer infrastructure
The Decision Layer runs entirely within your network. No SaaS dependency, no data exfiltration, no external telemetry tracking. Health data and return-to-work files never leave your network.
GDPR Article 9 by design
Health data architecturally separated. Return-to-work module with dedicated database, dedicated access circle, dedicated retention periods. Compatible with GDPR Article 17 and statutory retention requirements.
Data residency in detail →AI Act compliant
Clear architectural separation: the AI Agent classifies leave types and detects conflicts. Entitlement and deadline calculations run deterministically through rule engines. No black box in decisions.
EU AI Act readiness →Audit trail (append-only)
Signed decision records. Input hash plus rule version yields reproducible result. Sealed audit packages (JSON + PDF, SHA-256). Employee representatives can trace every decision.
ISO 27001 / SOC 2 cert-ready
Integrated controls registry, automated evidence runs, versioned policies. Compliance during operations, not documented after the fact.
Cert-Ready by Design →SSO & tenant isolation
Integration with existing identity providers. Tenant isolation at database level (Row Level Security). Return-to-work access circle separate from absence module. Legal entities cleanly separated.
More Agent Solutions
HR Agents are part of our agent portfolio. We offer specialized solutions for Finance and other domains.
HR & People Operations
All HR Agents
Payroll, Travel, Recruiting, Leave, Lifecycle, Screening. Works-council ready.
Finance & Accounting
Finance AI Agents
Document processing, account assignment, depreciation logic, correction postings.
Custom
Co-Build for Your Process
Compliance, operations, shared services - we develop agents in a co-build model.
Deep Dive in the Agent Briefing (Gosign Magazine)
Our expert article series for decision-makers deploying AI Agents in the enterprise.
Frequently Asked Questions about the Leave & Absence Decision Layer
What is the difference between the Leave & Absence Decision Layer and SAP SuccessFactors Time Off or Workday Absence?
SAP and Workday are data systems: they record absences, calculate balances, and generate approval workflows. The problem lies upstream: which policy applies? Which collective agreement takes precedence? Is the leave entitlement correctly prorated after a part-time change? The Leave Decision Layer makes these decisions rule-based and documents every single one. SAP and Workday remain your system of record. The Decision Layer becomes your system of decision.
Do I need to replace my existing absence system?
No. The Decision Layer sits between your source systems and your HRIS. It reads data from time management, employee master records, digital sick notes, and shift planning, makes decisions, and writes results back into SAP SuccessFactors, Workday, or your HRIS of choice.
How does the Decision Layer handle return-to-work processes after extended leave?
Return-to-work processes after extended leave require particular data sensitivity. The Decision Layer calculates the absence threshold (configurable per jurisdiction - e.g. 42 sick days rolling in some EU countries, 28 weeks in the UK) automatically and triggers the process. Health data is architecturally separated (dedicated database, dedicated access circle). The agent sees only whether a return-to-work obligation is triggered - no diagnoses, no sick note reasons in the absence module. GDPR Article 9 by design.
Is the system compatible with employee representation requirements?
Yes. Employee representation bodies (such as works councils in Germany, or staff committees across the EU) hold co-determination rights over leave policies under applicable employment law. The Leave Decision Layer makes the rule engine transparent, rejection reasons traceable, and reports pseudonymised. Anomaly detection is controllable via feature flag. Employee representatives can trace every leave decision via the audit portal.
How is health data protected?
Health data belongs to special categories of personal data (GDPR Article 9). The return-to-work module is architecturally separated: dedicated database, dedicated access circle, only visible to the designated team. The agent in the absence module sees no diagnoses. Return-to-work files are maintained separately from the personnel file. Retention periods are configurable.
What do the simulation results on this page mean?
We configured the Leave Decision Layer for four industries with realistic parameters: collective agreements, leave regulations, shift models, absence frequencies. The zero-touch rates show which proportion of absence transactions can be fully automated. (US: No federal mandatory leave law. FMLA provides 12 weeks unpaid. State laws vary significantly.) In an initial consultation, we calculate with your parameters.
How does a project start in practice?
In a 30-minute call, we clarify your parameters: collective agreements, absence types, system landscape, volume. From that, we produce a pilot proposal: one site, one collective agreement, the most frequent leave types. Typical pilot projects launch within 3 months. Extensions to additional sites and leave types run in parallel.
Does any data leave the organisation?
None. The Decision Layer runs entirely within your infrastructure. No SaaS dependency, no data exfiltration, no external telemetry tracking. Health data, leave accounts, and return-to-work files remain in your network.
Let us run the numbers.
30 minutes. Your collective agreements, your absence rules, your result. We configure the Decision Layer with your actual parameters and show you what adds up.