This year’s SQLBits followed the main trends of recent years. Most sessions focused on AI, new features and best practices in Fabric, as well as the new SQL Server. One specific topic of this year’s event, discussed mainly behind the scenes, was the number of pedestrian crossings in front of the congress centre where the conference took place. So if you’re curious about what was discussed at the conference, here is an overview of the most important takeaways.
The Future of Working with Data in the AI Era
Will AI Really Replace Us?
According to the speakers, companies that proudly laid off large numbers of people, claiming they would be replaced by agents, started hiring them back very quickly. Reportedly, Oracle, which announced at the beginning of the year that it had laid off 30,000 people, is now hiring 8,000 of them back.
Whether this is because people are still cheaper, or because certain human work simply cannot be replaced that easily. At the same time, it is becoming clear that an agent can cause massive damage not over a matter of days, but within seconds.
So How Should We Use AI?
Our goal should not be to replace employees with an LLM. It makes much more sense to look for “synergy” between AI and people. This can enable faster work at a higher quality, but it will not happen without human involvement.
At the same time, it is important to realise that an LLM is not the answer to everything. There is still a category of tasks where AI has no place, and where a classic Power App can solve the problem better and cheaper.
What About Security?
Less than 5% of projects actively address security when working with AI applications. And even when a company does deal with security, it is often surprised afterwards by how much information it has provided itself.
🔐 One final reminder: Data is not collected by the LLM, but by the application you use to communicate with the model.
Using Fabric in Everyday Work
Another major topic was how to work correctly with the individual components of Microsoft Fabric. Sessions covered topics such as:
🔹medallion architecture,
🔹lakehouses and Delta tables,
🔹capacity optimisation.
These sessions confirmed that we are heading in the right direction and keeping pace with global standards.
It is also worth mentioning that Fabric is CI/CD ready. This means there is nothing standing in the way of smooth and secure development.
New SQL Server Features
SQL Server 2025 also attracted a lot of attention. Interestingly, AI was not used in its development to the extent we might expect today, because the main development work had already been underway more than two years ago.
Its main new features and capabilities include:
🔹integration with AI and readiness of SQL Server as a database for AI applications,
🔹support for new features, such as working with JSON, storing vector data types, and the ability to call REST APIs directly from SQL,
🔹continued strong integration with Azure.
And What About Those Pedestrian Crossings?
There were eight pedestrian crossings in front of the congress centre, which we had to cross every day. As the crow flies, the pavement and the building were only about 100 metres apart, but thanks to the roundabout, the walk could easily take 15 minutes.
And that was also part of SQLBits — data, AI, Fabric, SQL Server, and a surprisingly intense walking workout.
Register for the one-day online workshop Chat with Your Data in a Day — model smarter and get answers faster.
Who we help become data-driven
