Urgent
⚖️
EU AI Act Compliance
Risk classification, FRIA assessments, transparency audits, and regulatory-grade documentation: tailored for SMEs who can't afford Big 4 pricing.
Risk ClassificationFRIADocumentation
EU AI Act enforces 2 August 2026 - are you ready?
We Make AI Explainable to Humans, Not Just Machines
Luxembourg-based consultancy combining Explainable AI engineering with sociological research. We audit your algorithms and test whether real people understand their decisions.
✓ EU AI Act Compliance
✓ XAI Expertise
✓ Sociological Research
Enforcement: 2 August 2026
The EU AI Act Changes Everything
High-risk AI systems such as hiring, credit scoring, insurance, public services, must demonstrate transparency, fairness, and human oversight. Penalties reach €35M or 7% of global revenue.
Most compliance providers audit algorithms. We also audit whether the humans affected by those algorithms understand them. That's what the AI Act actually requires and what nobody else delivers.
See how we help →-
Days until enforcement
65K+
High-risk AI systems in EU
€3.4B
Annual compliance market
0
XAI+Sociology firms in LU
Services
From AI Act compliance to algorithmic bias auditing: we cover the full spectrum of trustworthy AI, with the human dimension others miss.
🔍
Algorithmic Bias Auditing
Independent third-party audits for AI in HR, finance, insurance, and public services. Statistical fairness testing + sociological impact analysis.
Fairness MetricsProxy DetectionRemediation
👥
XAI User Studies
Do your AI explanations work for real users? We test comprehension, trust calibration, and actionability across demographics and languages.
Trust TestingComprehensionMultilingual
🎓
AI Literacy & Training
Article 4 of the AI Act mandates AI literacy for all staff. Workshops that non-technical teams actually retain which are co-designed by an AI expert and sociologist.
AI Act EssentialsHR TeamsWorks Councils
🏛️
AI Ombudsman (Public Sector)
Retained advisory for municipalities and government agencies. We review every AI system before deployment (citizen perception studies included).
RetainerFRIA for Public AIAnnual Reports
⚡
Explainability Design Sprint
One-week intensive: audit your AI, interview 10 users, identify gaps, co-design explanation interfaces, deliver an AI Act-ready specification.
5-Day SprintProduct TeamsStartups
Technical Services: Data Pipelines, NLP & Forecasting
AI Architecture & Audit
SHAP-based explanation pipelines for ML models. Regulatory-grade transparency reports for FinTech. Built on FNR Bridges research.
EU Grant Strategy
Full lifecycle support for FNR, Luxinnovation, and Horizon Europe proposals. Consortium coordination and impact reporting.
Energy & NLP Pipelines
Time-series forecasting, real-time sentiment engines, and custom Python/Django infrastructure for operational data.
Why Leonarc
Every competitor has data scientists. Nobody else has a sociologist. That's our edge.
1
We audit algorithms and their users
Competitors check the code. We also test whether a nurse, a loan officer, or a citizen actually understands what the AI told them.
2
Sociology is our moat
User studies, trust calibration, social acceptability testing, multilingual comprehension like sociological methods no SaaS platform can replicate.
3
Built on peer-reviewed research
Our methods come from FNR researches and publications.
4
Luxembourg-first, SME-sized pricing
First dedicated XAI + AI ethics consultancy in Luxembourg. Greater Region coverage at a fraction of Big 4 rates.
Typical auditor vs. Leonarc
Typical auditor:"Your SHAP values show feature X drives 40% of decisions"
Leonarc adds:"…and our user study shows 73% of loan officers misinterpret that explanation, creating a false sense of confidence. Here's how to redesign it."
Typical auditor:"Demographic parity gap is 4.2%, within tolerance"
Leonarc adds:"…but interviews with rejected applicants reveal they had no actionable recourse. The system is statistically fair yet socially opaque."
Sociology × AI
Sociological analysis translated into practical AI strategy: teams deploy faster, manage risk better, and earn stakeholder trust.
Behavioral Context in Model Design
We model real decision behavior, institutional constraints, and user workflows to improve model relevance, reduce friction at rollout, and increase operational accuracy.
Fairness, Power & Accountability
We identify who benefits, where bias surfaces, and which controls are needed (then convert that into governance frameworks for executives, auditors, and regulators).
Human-Centered Deployment
We align AI tools with team incentives, communication patterns, and stakeholder expectations to improve adoption, shorten implementation cycles, and sustain value.
Track Record
Applied research and delivery across Explainable AI and real-time NLP.
Explainable AI & FinTech
FNR
Transparent Credit Scoring & Classification
SHAP-driven explanation pipelines for auditable financial ML models. Methodology for Refined Classifications using prototypical explanations.
Financial NLP
Large-Scale Financial Text Analytics
Multi-class text classification for financial document collections. Pre-trained language models for automated categorisation at scale.
Let's Talk
Whether you're preparing for the AI Act, need a bias audit, or want to make your AI understandable to real people, we're here to help.
No pitch decks. No upselling. Just a clear assessment of what you need.
Contact Us