Skip to content

AI Transparency Statement (EU AI Act)

In accordance with the EU Artificial Intelligence Act (AI Act), we transparently inform you about the use of AI systems on this website.

Last updated: April 2026

AI Systems in Use

Portfolio Chatbot

Modell / Model
Mistral Small (via Scaleway Inference, EU)
Zweck / Purpose
Answering questions about services and projects
Risiko / Risk
Minimal Risk (Category: General Chatbot)
Daten / Data
No personal data is stored. Chat history only in the browser.

RAG Demo

Modell / Model
BGE-Gemma2 Embeddings + Mistral Small
Zweck / Purpose
Demonstration of Retrieval-Augmented Generation
Risiko / Risk
Minimal Risk (Category: Demo/Showcase)
Daten / Data
Uploaded documents are automatically deleted after 24 hours.

Code Playground

Modell / Model
Devstral Small (via Scaleway Inference, EU)
Zweck / Purpose
Code generation for demonstration purposes
Risiko / Risk
Minimal Risk
Daten / Data
No storage of generated content.

Anonymization Demo

Zweck / Purpose
Demonstration of PII detection and anonymization
Risiko / Risk
Minimal Risk
Daten / Data
All data is processed only in memory, not stored.

Contract Generation

Modell / Model
Mistral Small (via Scaleway Inference, EU)
Zweck / Purpose
AI-assisted creation of service contracts from project data. Generated drafts are ALWAYS reviewed and manually approved by the admin.
Risiko / Risk
Limited risk (Human-in-the-Loop: Admin approval before delivery)
Daten / Data
Project data (name, milestones, budget) is processed. No personal data is sent to the LLM.

Test Checklist Generation

Modell / Model
Mistral Small (via Scaleway Inference, EU)
Zweck / Purpose
Automatic creation of customer-friendly test checklists from technical task descriptions.
Risiko / Risk
Minimal risk (no decision-making function, suggestions only)
Daten / Data
Task titles and descriptions are processed. No personal data.

Skill Factory / Knowledge Base

Modell / Model
Mistral Small + BGE-Gemma2 Embeddings (Scaleway, EU)
Zweck / Purpose
Generation of project-specific development skills and knowledge management via RAG (Retrieval-Augmented Generation).
Risiko / Risk
Minimal risk (internal use only, no customer contact)
Daten / Data
Publicly available skill documentation. No personal data.

AI-Assisted Software Development

Modell / Model
Claude (Anthropic), Gemini (Google) — as development tools
Zweck / Purpose
Assistance with code development, code review, and documentation. All AI-generated code changes are logged in docs/history/ per EU AI Act.
Risiko / Risk
Limited risk (Human review: Every change is reviewed by the developer)
Daten / Data
Source code is processed. No customer data is sent to external AI systems.

EU Hosting

All AI models are operated via Scaleway Inference (Data center: Paris, France, EU). No data transfer to third countries. AI development tools (Claude, Gemini) process source code only, no customer data.

Human Oversight

AI-generated content on this website is labeled as such (badge: 'AI-generated'). All AI systems are operated under human oversight.

Audit Trail (Art. 26 EU AI Act)

All AI-assisted source code changes are logged in docs/history/ per EU AI Act Art. 26. Every Git commit with AI participation carries a Co-Authored-By header for traceability.

Ihre Rechte / Your Rights

You have the right not to be subject solely to automated decision-making. For questions about AI usage: info@schwankl.info

Made in Germany
100% GDPR Compliant
AI Act Ready
Secure Hosting
Accessible
Cookie Consent
Data Anonymization
KI-Transparenz | AI Transparency | Harald Schwankl