# moepinion > moepinion is the editorial blog of moe & company, a Swiss AI and data science > consultancy based in Winterthur, Zürich, Switzerland. Posts cover AI strategy, > data science, forecasting, software engineering practice, and change management > — written by the consultants who do the work. ## About moepinion is where the people at moe & company write about the things they think about: software, data, building products, and ideas worth turning over. It reflects the same expertise the consultancy applies for clients — AI strategy, data science, data engineering, MLOps, and organizational change management for AI adoption — in essay form rather than client deliverables. For consulting services, see the [main site](https://moecompany.com) (its [llms.txt](https://moecompany.com/llms.txt)). ## Articles - [The model won't tell you it's guessing](https://moepinion.moecompany.com/blog/the-model-wont-tell-you-its-guessing/): 2026-07-13, by Matteo Gasser — Most models answer with the same tone whether they're certain or not. If you care about reliability, you have to build the signal yourself. - [When your embeddings stop understanding your documents](https://moepinion.moecompany.com/blog/when-your-embeddings-stop-understanding-your-documents/): 2026-07-08, by Matteo Gasser — Embedding model choice feels like a one-time decision. It isn't. Domain language drifts, and your model doesn't follow it. - [What chasing the top-ranked model actually costs](https://moepinion.moecompany.com/blog/chasing-leaderboards-is-burning-your-ai-budget/): 2026-06-22, by Matteo Gasser — Moving to the new leaderboard leader every six months slowly eats most of the savings you wanted, and nobody tracks the bill. - [What a leaked system prompt says about yours](https://moepinion.moecompany.com/blog/leaked-prompts-reveal-your-own-prompt-hygiene-problem/): 2026-06-22, by Matteo Gasser — A leaked system prompt is a good reason to check if your own production prompts have an owner, a version history, or any review. - [The unglamorous part of sovereign AI](https://moepinion.moecompany.com/blog/sovereign-ai-dies-in-the-data-room/): 2026-06-17, by Matteo Gasser — National language-model projects usually stall on who owns the training text and who can allow its use, not on compute or architecture. - [What your AI vendor can change without telling you](https://moepinion.moecompany.com/blog/you-dont-control-the-ai-you-bought/): 2026-06-15, by Moe & Company — When you license a model, the vendor often keeps the right to change how it behaves, with little duty to tell you first. - [Most of our LLM failures were predictable](https://moepinion.moecompany.com/blog/your-llm-isnt-a-black-box-you-are/): 2026-06-08, by Matteo Gasser — When an LLM fails in production it is usually a known failure mode we did not plan for, not the model being mysterious. - [Approval buttons that don't show you the decision](https://moepinion.moecompany.com/blog/your-ai-agent-approval-button-is-security-theater/): 2026-06-01, by Matteo Gasser — Most AI agent approval dialogs hide the data, scope, and reversibility you need to judge the action before you click approve. - [What employees are really objecting to](https://moepinion.moecompany.com/blog/your-employees-are-booing-the-pitch-not-the-tool/): 2026-05-20, by Matteo Gasser — When teams resist a new AI rollout, the real problem is usually how it was announced, not the software itself. - [What If We're Wrong About What's Possible?](https://moepinion.moecompany.com/blog/what-if-were-wrong-about-whats-possible/): 2026-04-28, by Bart Thomson — Every industry has a 'that's impossible', until someone proves it isn't. What's yours? - [Der Nutzen von Change-Management: KI-Tools Mitarbeitenden näherbringen](https://moepinion.moecompany.com/blog/der-nutzen-von-change-management/): 2025-12-16, by Andreas Margelisch — Wie Change Management dabei helfen kann, Mitarbeitende für den Einsatz von KI-Tools zu befähigen – und was dabei wirklich zählt. - [In praise of boring software](https://moepinion.moecompany.com/blog/the-case-for-boring-software/): 2025-02-05, by Matteo Gasser — Why our most reliable systems run on PostgreSQL, Docker and Nginx, and what newer tools really cost you later. - [What a forecast can and can't tell you](https://moepinion.moecompany.com/blog/what-forecasts-cant-tell-you/): 2025-01-14, by Matteo Gasser — A forecast gives you a range of likely outcomes if the past keeps holding, and it goes quiet when the process changes. ## Topics Articles are tagged by subject. Browse by topic: - [AI & Automation](https://moepinion.moecompany.com/tags/ai-automation/): 10 articles - [Engineering](https://moepinion.moecompany.com/tags/engineering/): 7 articles - [Perspective](https://moepinion.moecompany.com/tags/perspective/): 3 articles - [Change Management](https://moepinion.moecompany.com/tags/change-management/): 2 articles - [Forecasting & Data](https://moepinion.moecompany.com/tags/forecasting-data/): 1 article Browse [all topics](https://moepinion.moecompany.com/tags/). ## Feed - [RSS feed](https://moepinion.moecompany.com/rss.xml) ## Keywords AI consulting Switzerland, data science consultancy Switzerland, AI strategy Zürich, machine learning consulting Switzerland, data engineering Switzerland, MLOps Switzerland, AI transformation consulting, change management AI, forecasting consulting Switzerland, moe & company blog, moepinion ## Contact - Blog: [moepinion.moecompany.com](https://moepinion.moecompany.com) - Main site: [moecompany.com](https://moecompany.com) - Email: [info@moecompany.com](mailto:info@moecompany.com) - LinkedIn: [moe & company](https://www.linkedin.com/company/moe-company) - Address: Stadthausstrasse 14, 8400 Winterthur, Switzerland - Legal name: Moe & Company AG (CHE-362.702.118)