We use artificial intelligence and machine learning to support cantonal administration employees and deliver better public services. We view AI as one of many tools for digital transformation.

Team Data at the Statistical Office of the Canton of Zurich operates as a data science competence center. We work on AI pilot projects for the cantonal administration, based on RRB 1331/2022, action area E4 of the Strategic Initiative Data, and legislative goals to expand the competent use of AI in administration.

Our pilot projects explore whether machine learning can solve specific problems in our business processes. Together with cantonal administration partners, we develop prototypes and proofs of concept to validate potential solutions. Learn more about how the canton works with AI.

We share experiences, code, and data from our pilot projects here on GitHub whenever possible. We welcome feedback—please contact us via email or open issues/PRs in the respective repositories.

• TranscriboZH Audio Transcription: Transcribe any audio or video file. Edit and view transcripts in a standalone HTML editor.
• Hybrid Search: Intelligent search application for large document collections.
• Document Research Tool: Perform intelligent research over document collections using hybrid search and LLMs.
• Deep Research: Powerful, automated research and analysis across your own document collections.
• AI Chat: A locally operated LLM chat with document processing capabilities.
• Semantic Search Evaluation Tool: A framework for evaluating semantic search across custom datasets, metrics, and embedding backends.
• Hybrid Search Evaluation Tool: A framework for benchmarking embedding models in hybrid search scenarios (BM25 + vector search). Measure MRR@K, Hit@K, embedding latency, and memory consumption. Bring your own data or use MTEB-compatible datasets.
• Named Entity Recognition (archived): NER framework tailored for administration use cases.

• ZIX Understandability Index: Measure how understandable a German text is.
• Language Simplification Tool: Simplify your institutional communication. Get rid of «Behördendeutsch».
• Language Simplification API: Language Simplification as an API.
• Glossarizer: Create plain language glossaries from texts, URLs, and files using LLMs.
• Alt Text Generator: Automagically generate German alt texts for web images using AI.
• Subtitle to Speech Tool: Convert subtitle files into synchronized German language audio.

• OGD AI Analyzer: Analyze the quality of a DCAT metadata catalog.
• OGD AI Metafairy: Easily create high-quality dataset descriptions.
• OGD AI Search: Search semantically, lexically, and multilingually in your OGD metadata catalog.

• Plausi App: Predict votes and detect anomalies using R.