Bridging AI Agents to Power BI & Using Jev with SQL Server

Together with inovex GmbH, we’re bringing the SQL Server community together again for an evening of technical talks, real-world experiences, and good conversations.
inovex is hosting us at their office and taking care of the refreshments.

Thank you for supporting the community and making the evening possible.

Come by, learn something new, ask questions, meet other data people, and stay for the conversations afterward.

6:00 pm - Open door
6:30 pm - First talk & Q&A
7:15 pm - Break with Snacks & Networking
7:45 pm - Second talk & Q&A
8:30 pm - Wrap-up and Time for More Conversations
9:00 pm - Close door

First talk: Bridging AI Agents to Power BI (Tobias Maasland)
As AI-assisted development tools like Claude Code and Gemini CLI mature, running them inside isolated Linux container sandboxes (e.g., Podman) has become a security standard. However, accessing native Windows capabilities - such as Microsoft’s Power BI Modeling MCP server - presents architectural challenges due to stdio limitations across OS boundaries, and Hyper-V firewall restrictions.
In this session, we examine these cross-platform friction points and demonstrate how a lightweight proxy bridge enables sandboxed CLI agents to seamlessly inspect models and execute DAX queries against local Power BI Desktop instances without sacrificing container isolation.

Tobias Maasland is a Data Product Owner and Data Engineer focused on building value-driven data solutions and intelligent workflows. With strong expertise in cloud ETL/ELT, data warehousing, and rapid AI prototyping, he actively incorporates Google Gemini and agentic AI systems directly into the data domain. Tobias specializes in turning initial PoCs into robust, end-to-end data products through structured requirement analysis and AI-assisted engineering.

Second Talk: Using Jev with SQL Server (Sascha Lorenz)
Jev takes a different approach to AI: instead of generating text, it makes fast, structured decisions based on the data you provide.
In this session, we will take a quick look at what Jev is, how its decision model differs from traditional LLMs, and where this approach can be useful when working with SQL Server.
We will then get practical: starting with a simple Jev API call and moving on to calling Jev directly from SQL Server 2025 using its new external REST endpoint capabilities. Along the way, we will explore a few examples of using database data for classification, scoring, and decision-making, and also discuss where this belongs in a real SQL Server architecture and where it probably does not.

Sascha Lorenz has been working with SQL Server and enterprise systems for more than 30 years. As founder of PSG Projekt Service GmbH and former Microsoft MVP, he focuses on performance, troubleshooting, automation, and building better tools for understanding real-world database workloads. He is especially interested in what happens when traditional database engineering meets observability, AI, and new developer tooling.