With AI agents, Power BI custom visuals take days instead of weeks – provided there is a plan. Using a radar chart from a customer project, Michael Wetzel shows why the specification is half the work and how Performance Analyzer narrows down bugs.
Could time series foundation models improve the forecasting of district heating demand? In our offline study, we compared Chronos-2, TimesFM-2.5, and Moirai-2 with AutoARIMA, XGBoost, and other baselines across four real district heating grids. Chronos-2 produced the most accurate results, even in zero-shot settings without target-grid training data. We also examine data efficiency, fine-tuning, resilience to faulty sensor inputs, and operational reliability.
Clarity often only emerges when planning data products or software projects once the development phases have been clearly defined.
Taking a step-by-step approach can gradually reduce uncertainty and align expectations within interdisciplinary teams.
AI image generation has made a massive leap forward since 2025, evolving from distorted results to the production of photorealistic 4K images in real time. In this article, we explore how modern diffusion models work, offer practical tips for structured prompting and demonstrate how to achieve consistency across multiple images using a customer project on storyboard creation. We also explain which models are currently the best and where the technology’s limits lie.
Data science, AI and cloud architectures form the core of our business, but it’s sometimes beneficial to step outside your comfort zone. That’s precisely what our colleague sat scieneers did in mid-December at IT Days 2025 in Frankfurt. We took away ideas from topics such as software architecture, DevOps, agile methods, and digital sovereignty that will directly influence our daily work on scalable RAG systems, clean software architecture, and monitoring.
Once again, the team took centre stage. All scieneers from Karlsruhe, Cologne and Hamburg came together for a two-day autumn event at the end of September. As well as having some exciting discussions and taking part in some joint activities, we also welcomed five new colleagues. We are now a team of around 50 people!
PyData Berlin 2025 at the Berlin Congress Center was three days full of talks, tutorials, and tech community spirit. The focus was on open-source tools and agentic AI, as well as addressing the question: How can LLMs be used productively and in a controlled manner? We from scieneers gave a presentation on LiteLLM, titled “One API to Rule Them All? LiteLLM in Production”.
In our ‘Bring Your Own Data’ workshop, you will learn about the capabilities of Microsoft Fabric, from connecting data to integrating AI, all with your own data. It is ideal for business users, data professionals, and decision-makers who are looking for a practical introduction to modern data platforms.
With our ‘Data Science for Everyday Work’ training course, we aim to help companies bridge the gap in practical knowledge when implementing machine learning projects and foster a shared understanding within the team.
We also offer individual modules that can be combined as desired, and we provide consultancy services for your specific data challenges.
Most of our colleagues from our three locations Karlsruhe, Cologne, and Hamburg met in Hamburg for two days. We discussed specialist and internal topics, gained new ideas, and shared experiences.