This summary is produced by the author, and not by AI.
Power BI DC Days is a free community event focused on Power BI and Microsoft Fabric. If there was a theme of the event, it was "fighting against fragmentation".
At the very beginning of the keynote, the organizers shared that the prior year they had dealt with a lot of burnout. One of their goals this year was to avoid that, and they had even considered cancelling the event as a result. In my own experience as an event organizer, it simply feels more challenging to run an event or user group than it did 8-10 years ago. Venues, sponsors, and attendees are harder to find.
Instead of cancelling the event, they expanded the organizing team and spread out the work. The reason that they kept the event was an interesting one: as a result of their location, Power BI DC Days is one of the few events that focuses on work in sovereign clouds (such as the US Federal Government's GCC environment) where Microsoft Fabric is not yet available. In a very real sense, this group was meeting needs very specific to their community.
Another piece that was interesting is that Power BI DC is part of a larger collective of 8 user groups, Data Community DC. While I've seen overlap before with SQL Server and Power BI user groups, this was the first time I had seen so many diverse user groups working together.
During the event, I attended 4 sessions that touched on AI and I gave one of my own. I was very curious how much we all are still trying to figure out the best way to make use of AI. The range of topics, and more importantly what people took notes on, was incredibly wide. My general sense is that as a community we are still trying to cut through the hype and establish a set of recommended practices.
You had sessions covering things as simple as adding Power Query comments with GitHub Copilot or what a SKILL.md file is. The most complex session that I attended covered how the Azure Data Insights & Analytics team at Microsoft build on Fabric and how their AI use evolved over time. Their approach to enabling AI was a complex one that leaned heavily on CI/CD practices and being willing to change course when the AI wasn't providing the right results.
There were repeating themes of investing in the quality of your semantic model, improving comments and descriptions, and making use of the PBIP format and Git so you can safely make changes. If you are uncertain about how to apply AI to Power BI, focusing first on these fundamentals is a good starting point no matter what.
At no point was AI presented as a magic button, other than for small, minor tasks like adding comments to Power Query. Even then the instructions and prompting were presented as an interactive process, to avoid issues like the LLM putting pieces of code into the comments themselves. Anyone planning on working with AI should be prepared for it to be an iterative process.
Fragmentation of the tools available and of AI maturity can make it difficult to solve problems and get oriented. When providing guidance on implementing solutions, it's important to be careful in assuming what tools will be available. While maintaining a community is consistently challenging, sharing real experiences is one way to help with this issue of fragmentation and marketing hype.
See how Tabular Editor helps you build on the fundamentals that make AI adoption safer.
Give Tabular Editor a spin