HR & People Operations
Sick leave, onboarding, employment references, contract review, policy queries. Works council-ready, EU AI Act compliant.
View HR AgentWhen AI makes decisions – who is accountable?
The Decision Layer makes every AI decision traceable, auditable, and works council-ready.
AI agents can process sick leave, classify invoices, and review contracts. But who decides what the agent may do on its own and where a human must step in? The Decision Layer defines this – for every single process step.
Why AI Projects in HR Fail
HR processes depend on the knowledge of individual employees. Who knows which special leave policy applies at which location? Who remembers the difference between the company agreement from 2019 and the updated version from 2024? Who checks whether sick leave was correctly validated against the collective agreement?
This knowledge lives in people's heads, in email threads, in folders nobody can find. When someone leaves the team, the knowledge leaves too.
AI can help – but only when it's clear which rules apply. And who is ultimately responsible.
What Is the Decision Layer?
The Decision Layer is the governance layer between AI agent and target system. It decomposes every business process into individual decision steps and defines for each step upfront: Does a human decide, a rule set, or the AI? Every decision is automatically documented – for auditors, works councils (Betriebsrat), and internal audit.
The Decision Layer is not an AI agent – it's the governance layer above. It complements existing systems like SAP SuccessFactors, Workday, or DATEV and controls what an AI agent may do with these systems.
How the Decision Layer addresses shadow AI and technically enforces company agreements is detailed in Article 6 of the Blueprint 2026.
| Without Decision Layer | With Decision Layer | |
|---|---|---|
| Who decides? | Unclear – the agent delivers a result | Defined per step: human, rule set, or AI |
| Company agreements | Manually followed – or forgotten | Stored as fixed rules, technically enforced |
| Traceability | Result visible, decision path not | Complete documentation per decision |
| Auditor | Must manually review each case | Direct access to decision documentation |
| Works council | Blocks – no transparency | Supports – every decision traceable |
How Does the Decision Layer Work in Practice?
Sick leave example: 6 steps, clear accountability at each step. The Decision Layer defines for each: rule set, human, or automatic.
Why Do AI Projects Fail at the Works Council?
The most common reason AI projects fail in German enterprises: The works council (Betriebsrat) blocks them. Not because they oppose technology – but because they lack transparency. The Decision Layer solves this:
✓
Every company agreement is stored as a fixed rule. The agent cannot bypass it.
✓
For decisions affecting employees, a human always decides. Technically enforced, not just agreed upon.
✓
Every AI decision is documented: What was checked, which rule applied, what was the result.
✓
The works council can trace how any decision was made, at any time.
The difference: Others promise transparency. The Decision Layer enforces it technically.
How Does an AI Decision Become Audit-Proof?
Your auditor sees exactly what happened.
Which document was processed – and when?
Which rule was applied – and in which version?
How confident was the agent in its assessment?
Did a human review – and if so, who?
What was the result and when was it booked?
Which Decisions Can AI Make on Its Own?
Some decisions an AI agent can make on its own. Others need human review. And for strategic questions, the agent only provides data. The Decision Layer defines this – per step, not per process.
Who Is the Decision Layer For?
Head of HR / CHRO
You want to use AI in HR – without losing control. The Decision Layer ensures company agreements are enforced, the works council has transparency, and every decision is traceable.
CFO / Head of Finance
Every AI-assisted booking is audit-proof. Your auditor sees the complete decision path. Correction bookings are reduced because rule sets are applied consistently.
Works Council (Betriebsrat)
No black box. Company agreements are technically stored and cannot be bypassed. For personnel decisions, a human always intervenes.
IT / CTO
Model-agnostic, infrastructure-agnostic, no vendor lock-in. Technical details in the Reference Architecture →
The Decision Layer for Your Processes
Finance & Payroll
Document processing, account assignment, payroll, closing entries. Audit-proof. DATEV and SAP integration.
View Finance AgentDocument Processing
Invoices, contracts, certificates. Automatic classification and data extraction with rule-based validation.
View Document AgentsIs the Decision Layer EU AI Act Compliant?
The EU AI Act requires transparency, human oversight, and documentation of AI decisions. The Decision Layer addresses these requirements as an architectural principle – not as an afterthought compliance project.
Who Builds the Decision Layer?
The Decision Layer is developed and implemented by Gosign GmbH. Gosign is an Enterprise AI Infrastructure & Agent Engineering company based in Hamburg, Germany, with over 20 years of experience building complex systems for enterprises including Airbus, Deutsche Telekom, and Sony Music.
4–6 weeks to the first productive process. Full code transfer, no ongoing license fees. Goal: After 12–18 months, you operate your agents independently.
Frequently Asked Questions About the Decision Layer
What is a Decision Layer – simply explained?
The Decision Layer is the governance layer between AI agent and target system. It works like a standard operating procedure with clear decision stages – except it's technically enforced, not written on paper. For every step in a business process, it's defined upfront: Does a human decide, a rule set, or the AI?
Do I need to replace existing HR systems?
No. The Decision Layer complements SAP SuccessFactors, Workday, Personio, or DATEV. It controls what the AI agent may do with these systems – and documents every interaction. No migration, no system replacement.
How does the works council react to AI agents with a Decision Layer?
Typically positively – because the Decision Layer provides exactly the transparency that works councils (Betriebsrat) demand. Company agreements become technical rules, personnel decisions stay with humans, and every AI decision is traceable.
What distinguishes the Decision Layer from SAP Joule or Microsoft Copilot?
SAP Joule and Microsoft Copilot are AI agents – they execute tasks. The Decision Layer is not an agent. It's the governance layer above: it defines which decisions an agent may make autonomously, where a human must step in, and where rule sets apply. The Decision Layer is model- and vendor-agnostic.
How long does implementation take?
4–6 weeks to the first productive process. Week 1: Understanding your processes and rule sets. Weeks 2–5: Building and testing the first agent with Decision Layer. After that: Your team gradually takes over operations.
Deep Dive
Enterprise AI Infrastructure Blueprint 2026
The Decision Layer is the governance architecture that makes AI deployment works-council-ready, auditable, and scalable. Background on shadow AI, data classification, and the EU AI Act.
Which HR process costs you the most time?
Show us a specific process – we'll show you what the Decision Layer looks like for it.
Book a Meeting