Top 5 Takeaways from Datacon Seattle

As a Director in the Data and Analytics practice at Apps Associates, I’m always looking for opportunities to learn about the newest technologies and strategies available for our customers. As an eight-time Microsoft Data Platform MVP, I was fortunate enough to lead a full-day workshop and a few breakout sessions at Datacon in Seattle, WA June 23-27. These opportunities mean a lot to me, as they allow me to chat with attendees about our shared interests, challenges, and opportunities. Regardless of the buzzword of the moment, the most meaningful trends are the ones that real, hands-on teams are seeing in the real world.

Here are my top five takeaways from Datacon 2025:

Things Are Moving As Quickly As You Think

It’s not just you – the pace of software development, AI model development, releases, etc. is fast. For example, I presented a session on Data Agents in Microsoft Fabric, except the name changed back at the end of May and made my alliterative session title completely irrelevant. This speed isn’t likely to slow down anytime soon, which makes it all the more critical to find trusted sources to cut through the hype and keep your company focused on the key elements of your data estate as it modernizes. Whether that trusted source is an external partner or an internal team, it is OK, essential even, to ask for help keeping up with the speed of innovation in the data platform space these days.

Event-Driven Architecture for Data Is Here

Transactions are events. That simple sentence started a lot of interesting discussions at Datacon. While many organizations think of data movement in terms of traditional ETL and batch loads, it’s time to think about the data in your organization as data in motion. While some of that data is traveling quickly and some is not, realizing that a transaction is an event and the changing data in a database can be considered a data stream opens up new opportunities for efficiency in your data movement architecture. Real-Time Intelligence in Microsoft Fabric allows us to build these types of architectures in a low-code environment, further expanding the possibilities for efficient data movement that doesn’t rely on scheduled batch jobs. Taking automated actions on that data in motion using Activator allows us to start to think about data workflows in an “agentic” way, which also happens to be an ideal segue into our next takeaway.

Agentic AI Is Interesting To Most

Agentic AI was a common topic during my time at Datacon, even outside the session I gave on Data Agents themselves. There is a lot of buzz around this topic; a lot of the discussions I had with practitioners sounded something like this: “Our management team wants us to use Agentic AI – what is it and how can we do that?”.

Because the term is very popular right now, I’ve seen examples described as agentic that aren’t. It’s up to us as data professionals to communicate clearly about what we’re doing and what it is. For those of us focused on the Microsoft Data Platform, the agentic play is to integrate Data Agents in Microsoft Fabric with Copilot Studio (a new ability released in preview last week).

Want to learn more about how, and why, to do that? Contact us for a discussion that cuts through the marketing hype and lays out what this capability can and can’t do for you and your data.

We Need to Understand What AI Is – and Isn’t

As I noted in the last takeaway, the word “agentic” is overused. “AI” has been overused and misused for years now, becoming more of a ubiquitous marketing term in people’s minds rather than the powerful set of technologies it is. When stakeholders are demanding the use of AI in our products, in our dataflows, and in our decision making, it is all the more critical to understand what AI is and isn’t. There are places in our data platform where we are likely using “real” AI, like Copilot and chat with your data experiences. There are places where we are using a logical construct like an IF-THEN that is not AI – no matter how much people may want it to be.

Data Is Still Critical Infrastructure

Finally, no matter how much we’re talking about agents or AI or the next big buzzword, data remains critical to the success of any AI or AI-adjacent initiative. You can use the latest models to create a dazzling presentation, but if the data you are using isn’t clean and accurate, all you’re doing is a magic trick. If you drive a car blindfolded, you’re not taking actions based on all the correct information available to you, and you’re likely to take a destructive action without meaning to do so. If you allow agentic systems to take actions blindfolded, i.e. without clean, correct data available to them, they will also take destructive actions. Proper data architecture and engineering practices will never go out of style, no matter how much the data-driven world may change around them.

At Apps, we’ve been writing about and working with Azure AI for almost 10 years. We can help you cut through the hype of what AI is and isn’t around your data and build an AI-powered solution that will deliver real efficiency in your everyday processes.

Start the conversation with our team today to get started on your real AI roadmap.