Earlier in December, we gave two talks and shared the Tower story at the booths of two Python data events in Boston and Warsaw. Before the holidays, we’d like to share a few key takeaways.

Boston and Warsaw each highlight a different side of the same problem

Boston is a major center for data engineering research and new ideas. There’s a lot going on, and real progress is happening.

Warsaw has a strong and growing community of data engineers. Many try using big data vendor tools, but often find these platforms are more than they need for their actual tasks.

Using these big platforms well is also a real challenge. How do you organize data from different sources, manage your pipelines, and keep everything running smoothly?

The problem is that many people don’t even ask how. They just accept things as they are and try to manage. But there are ways to organize pipelines and make processing data easier. Tower is one of those ways.

Talking about experimenting with Data Agents really resonated with people

At both events we gave a talk on using Small Language Models to navigate the current Agentic AI hype. The rooms were packed, showing that a lot of people want to make the leap into the next generation of AI-driven data engineering.

AI is indeed everywhere these days, so it’s no surprise people are interested. It’s clear that many want to learn how to deploy data agents and get them to do useful, meaningful work.

One thing was clear from our conversations at the events: data is still at the core of every AI project. Without reliable data systems, nothing else works.

Data teams need to move quickly, and need reliable data infrastructure

This is probably the second biggest takeaway. Teams of any size need to move fast. When pipelines break or infrastructure fails, work comes to a halt. For some organizations, everything stops.

Manual fixes can hold things together for a while, but they don’t scale or help teams work well together. Without good management and automation, things get harder. Having engineers spend hours just watching pipelines run isn’t a real solution.

Your systems should grow with your pipelines, make teamwork easy, and let engineers focus on what matters most.

Thank you!

Thanks to NumFocus, the event organizers, and everyone who stopped by to see us. Let’s keep the conversation going on our Discord server.

We learned a lot and can’t wait to meet you at a future Python data event.