I'm writing this sitting in a hotel room. PASS Data Community Summit has just finished (and we're going to do it again next year, twice). Before writing editorials, unless I've got a burning idea, I like to glance back through the editorials I've written to be sure I'm not repeating myself too much. Funny enough, I published an editorial a year ago, immediately after the last Summit, emphasizing the need for human connection. I stand by that one of course, but I want to assure you I'm going in a different direction this time.

At the Summit this year, and I'm sure you'll be amazed and surprised by this, there was a lot of talk about Agentic AI. Microsoft's keynote had some. Redgate's keynote had some. There were a bunch of sessions from vendors and community members, all talking about Agentic AI. Whether we're talking about automating tasks, speeding software delivery, or making things like query tuning easier, everyone is planning to, as my friend Kellyn Gorman puts it, "rub a little AI on it."

Don't get me wrong, I'm into it. I love what's going on with Agentic AI. No, I don't think it belongs everywhere. My fridge does its job just fine without an agent. Yes, I'm aware of the breaches and other issues that have come up around all the agents everywhere. Sorry, but I don't care. I see the utility this new tool brings us.

However, one of the recurring themes you'll see from the people who are truly neck deep in this stuff is the importance of keeping humans in the loop. Judgement is the biggest reason. Yeah, sure, dropping the database and recreating it is the fastest way to deploy code. Several AIs have done this now. I think it's my fault because I used to joke about it so much. Having a human approve plans before they go to production, that's huge. However, it goes beyond that, way beyond.

I was asked by a young person at the event (and yeah, we had a ton of people without a single gray hair on their head, I'm so excited) whether or not it was even worth it for her to study databases, internals, coding best practices and all the rest. After all, the AI can handle that. My answer was an unequivocal yes. Want to know why? Because if you know how things work, you can build superior prompts for the AI. You can better review what the AI is doing. Building your knowledge, builds your judgement. And as we've already said, judgement is the biggest, most important reason to always have a human in the loop.

And one more on judgement, learning, and why people matter. One of the best ways to learn is at a live, in-person, event. You get to ask questions. You get to drill down. You get to share with others. You get to expand both your own knowledge base, and your network, which is another form of knowledge and judgement (never hurts to phone a friend).

So yeah, even in our Agentic AI Age of Wonders, people matter.