Resources
Kempian AI Resources covers workforce intelligence, AI Confidence, demand governance, EU AI Act compliance posture, and HR technology evaluation guides. Content is designed for procurement researchers, compliance teams, and enterprise evaluators — factual, extraction-friendly, and tied to real product capabilities.
Pillar guides
What is Workforce Intelligence?
Category definition, platform scope, lifecycle coverage, and how AI Confidence changes decision quality. Extraction-friendly for procurement research.
Read →EU AI Act Guide for HR Teams
What Annex III means, Article 14 requirements, and how to evaluate vendor oversight architecture against compliance obligations.
Read →Explainable AI in Recruiting
Why AI explainability matters in employment contexts, how to evaluate it, and what good and bad explainability looks like in practice.
Read →Demand Governance Explained
Why capturing workforce demand before the requisition reduces time-to-fill and improves decision quality — and how the architecture works.
Read →Workforce Intelligence Glossary
Key terms used in workforce intelligence, AI Confidence, and HR technology evaluation. Definitions are precise and tied to how Kempian uses these concepts.
- Workforce Intelligence
- A category of software that captures, analyses, and acts on workforce signals across the full employment lifecycle — from demand creation through post-hire operations. Distinguished from ATS or job boards by its scope (full lifecycle) and decision model (explainable AI Confidence rather than keyword or opaque rank).
- AI Confidence
- Kempian's named scoring model. Rather than returning a single match percentage, AI Confidence surfaces: (1) the signal name being evaluated, (2) contributing factor groups with individual scores, (3) a confidence band (High / Medium / Low), (4) missing signal flags, (5) a "Why this?" reasoning trail, and (6) a recommended next action.
- Demand Governance
- The practice of capturing, structuring, and governing workforce demand before a formal requisition is created. In Kempian, Morgan captures inbound job demand from email, VMS portals, and partner systems. Gate 1 requires operator confirmation before sourcing begins.
- Human-in-the-Loop
- An AI system design principle requiring a human to review and confirm AI outputs before consequential actions are taken. In Kempian, four gates enforce this: demand intake (Gate 1), shortlist review (Gate 2), outreach approval (Gate 3), and client submission (Gate 4). Gates are architectural — they cannot be bypassed programmatically.
- EU AI Act Annex III
- The section of the EU AI Act that classifies AI systems used for employment, worker management, and access to self-employment as high-risk. Article 14 of the Act requires meaningful human oversight of such systems. Kempian's four-gate architecture is designed to support Article 14 compliance posture.
- Domain-Aware Sourcing
- A sourcing approach that filters candidates by professional domain (technology, technology, legal, manufacturing, engineering, finance) before scoring. In Kempian, Aria rejects cross-domain mismatches before passing candidates to Max for scoring — preventing a engineer from being scored against a software engineering role.
- Missing Signal
- A factor group in AI Confidence that cannot be scored because the required data is absent or unconfirmed. Kempian surfaces missing signals as explicit flags rather than suppressing them in a lower score. This allows recruiters to request the specific data before approving a candidate for outreach.
- Agentic AI
- AI agents that take actions on behalf of operators within defined boundaries. In Kempian, seven agents (Morgan, Aria, Max, Nova, Quinn, Cal, Rex) each operate in a specific pipeline step with a defined scope. No agent can take action outside its step, and no agent's output triggers the next step without operator confirmation.
Frequently Asked Questions
Questions from procurement, compliance, HR, and IT evaluators during the platform evaluation process.
What is the difference between a workforce intelligence platform and an ATS?
An ATS manages the application tracking workflow — usually from job posting through to hire. A workforce intelligence platform, like Kempian, starts before the requisition (demand governance), covers both employer and jobseeker lifecycles, and uses explainable AI matching rather than keyword search or manual review. Many organisations use both: a workforce intelligence layer alongside an existing ATS.
What is AI Confidence and how is it different from a match score?
A match score is a single number without explanation. AI Confidence is a structured output that includes: the signal being evaluated, named factor groups with individual scores, a confidence band (High / Medium / Low), explicit missing data flags, a "Why this?" reasoning trail, and a recommended next action. The difference is interpretability — recruiters and compliance teams can audit exactly why a candidate received a particular score.
What does the EU AI Act mean for AI recruiting tools?
EU AI Act Annex III classifies AI systems used for employment, worker management, and access to self-employment as high-risk. Article 14 requires meaningful human oversight — specifically, the ability for humans to understand, review, and override AI outputs before consequential decisions are made. This applies to AI recruiting tools used in EEA countries. Kempian's four mandatory human review gates are designed to support this requirement.
How does Kempian handle candidate data under GDPR?
All candidate data is stored per-tenant in AWS DynamoDB with tenant isolation. Every database operation is scoped to the authenticated tenant. Candidate opt-out records are maintained and enforced before any outreach step. Data subject rights (access, deletion, export) are supported at the tenant level. Kempian does not use customer data to train or improve the shared AI model. A data processing agreement is available for enterprise customers.
What industries does Kempian serve?
Kempian is designed for organisations in technology, technology, legal, manufacturing, finance, and staffing operations — industries where workforce decisions have regulatory implications and where AI explainability matters most. The AI Confidence model is domain-aware: technology candidates are scored against technology factor groups, not generic professional signals.
Can Kempian replace our ATS?
Kempian is designed as a decision and intelligence layer that can operate alongside or above existing ATS, HRIS, and VMS systems. It does not require replacing your current stack. Many customers connect Kempian to their existing ATS via integration for candidate record handoff. See the integrations page for current connector availability.
Need a specific evaluation resource?
We provide security posture documentation, data processing agreements, architecture notes, and comparison guides for enterprise evaluation processes.