6 HR Processes. One Decision Layer. Full Source Code Access.
Payroll. Travel Expenses. Recruiting. Leave & Absence. Lifecycle. Candidate Screening.
Most HR processes rely on implicit knowledge held by individual employees - inconsistency across locations, audit risk during external reviews, know-how loss through turnover. AI Agents decompose every HR process into individual micro-decisions. For each step, it is defined in advance: does a human decide, a rule set, or the AI?
A selection from over 5,000 projects in 25 years of software development
HR Agent Portfolio
An HR Agent does not automate wholesale - it decomposes processes into micro-decisions. For each decision, the Decision Layer defines: human, rule set, or AI.
Payroll Decision Layer
Collective agreements, shift allowances, special payments. 4 industry simulations.
82-92% Zero-Touch
Travel Decision Layer
Every expense report decomposed into 5-12 auditable decision steps.
>80% cost reduction per receipt
Recruiting Agent
Screening, scheduling, standard communication. EU AI Act high-risk.
70-75% automation
Leave & Absence Decision Layer
Leave entitlements, absence rules, return-to-work obligations. 4 industry simulations.
78-87% Zero-Touch
Employee Lifecycle Decision Layer
Onboarding, internal transfers, offboarding in one auditable pipeline.
Candidate Screening Decision Layer
EU AI Act high-risk. Bias monitoring. Co-determination (Mitbestimmung) for selection criteria.
Why you should start with boring processes
The most effective HR agents don't automate the most spectacular processes - but the most repetitive ones. Payroll, time tracking, standard absences: high rule density, low risk, measurable ROI from day 1. The governance infrastructure you build along the way is the prerequisite for more complex agents.
The EU AI Act high-risk classification for HR AI systems stands. Under current law the obligations apply from 2 August 2026, but 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). The classification is not changing, so the time should be used to prepare. Starting with boring processes is the fastest path to building that governance infrastructure in production.
Free eBook: AI in HR
EU AI Act checklist, Decision Framework, works council perspective, and readiness assessment - the governance handbook for HR leaders.
Download for freeHow an HR Agent Works
Every HR process consists of dozens of individual decisions. The agent checks against all collective agreement criteria - more consistently than any specialist. Where employee representation bodies, employment law, or discrimination risk require it, a human decides.
Sick Leave Processing
| Micro-Decision | Who Decides | Why |
|---|---|---|
| Check continued pay period | AI autonomous | Checks against ALL collective agreement criteria - more consistent than any specialist, even on Friday at 4 PM |
| Assess return-to-work obligation (> 6 weeks in 12 months) | Human | Employee representation rights require human-in-the-loop (in Germany: works council co-determination), discrimination risk with health data |
| Inform line manager about absence | AI autonomous | Consistent information - absence and duration only, no diagnosis |
Onboarding
| Micro-Decision | Who Decides | Why |
|---|---|---|
| Pay grade classification | Human | Employee representation bodies have oversight rights (in Germany: co-determination (Mitbestimmung) under Works Constitution Act) |
| IT access and permissions | Rule set + AI | Standard permissions automatic, special access escalated |
| Create onboarding plan | AI + Human | AI creates proposal, hiring manager reviews and adjusts |
Employment References
| Micro-Decision | Who Decides | Why |
|---|---|---|
| Generate role description | AI | Based on job description, proposal for review |
| Formulate performance assessment | Human | Individual assessment, manager is accountable |
| Check legal compliance of language | AI + Human | AI checks against known patterns and flags risks, human decides final |
The pattern: Routine and rules are automated by the agent. Discretionary and legally sensitive decisions stay with humans. Every step is documented.
Decision Layer: Every Decision Auditable
The Decision Layer decomposes every HR process into individual decision steps and defines for each step: does a human decide, a rule set, or the AI? Where the model can decide with confidence, it decides autonomously. Where bias risk, discrimination potential, or co-determination (Mitbestimmung) is involved, the architecture enforces human review.
Input → Model → Assessment → Confidence Score → Reasoning → Decision Path → Outcome
Every decision produces a complete data record: input, rule, version, confidence, outcome, timestamp. Auditable for employee representation bodies, external auditors, and internal compliance.
Governance & Employee Representation
Human-in-the-Loop as Architectural Principle
Not as an optional approval loop, but as architecturally enforced routing. Autonomous decision at high confidence and low risk. Human-in-the-Loop for bias risk, discrimination potential, and employee oversight matters. Works agreements and collective agreements are mapped as explicit constraints in the Decision Layer - the agent cannot bypass them.
Audit Trail
Every agent decision produces a complete decision record: input, rule, version, confidence, decision path, outcome, timestamp. Employee representation bodies can trace all decisions through the Auditor Portal. How employee representation bodies shift from blockers to enablers is covered in Article 9 of the Blueprint 2026.
EU AI Act Readiness
The architecture addresses transparency (Art. 13), human oversight (Art. 14), recording obligations (Art. 12), and risk management (Art. 9) as a design principle - no retroactive compliance project needed. (UK: Post-Brexit, the UK pursues a sector-specific approach rather than a single AI Act.)
Integration into Your Existing System Landscape
AI Agents do not replace systems. SAP SuccessFactors stays your HCM, Workday stays your HR platform, SAP FI/CO stays your ERP. Agent logic is decoupled from the target system - via standardised interfaces. No migration project, no system change. One process, one system, one agent.
From PoC to Platform
Discover - 1 week
Process analysis, understand rule sets, prioritise use cases. Which process has the highest error potential?
Build - 3-4 weeks
Production PoC. One agent, one process, live in your infrastructure with Decision Layer and Audit Trail.
Scale - Ongoing
More agents, more processes. Same governance, same auditability. After 12-18 months, you operate your agents independently.
HR Agent Readiness Assessment
7 questions, 3 minutes: How ready is your HR organisation for AI Agents?
Start assessment →HR Agent Catalog: 48 Agents Assessed and Prioritized
Which HR agents exist, what governance do they need, and in what order should you start? The catalog assesses 48 agents across 6 dimensions - from Agent Readiness to EU AI Act classification.
11 domains, 3 interactive prioritization quadrants, micro-decision tables for every agent.
Explore the HR Agent Catalog →Deep Dive in the Agent Briefing (Gosign Magazine)
Frequently Asked Questions about HR Agents
Which HR products does Gosign offer?
Is this compatible with works councils (Betriebsrat) and employee representation bodies?
How long does a pilot project take?
Does the HR Agent replace the HR department?
How does the HR Agent comply with the EU AI Act?
Where does the AI decide better than a human?
Talk to us about a specific HR process.
Whether payroll, travel expenses, recruiting, or onboarding - we start with one process, one agent, one week of discovery.