Machine Learning: Your Product’s Superpower and Worst Nightmare

Hey, Valencia product people! 👋

Our October ProductTank event is ready, and you’re invited!
This time, we’re exploring machine learning for product people, with an expert speaker to help us make sense of the topic and what it means for our work.

📍 New venue: We’ll be hosted by Deutsche Telekom this time, so make sure you head to the right location!

About the talk

Machine learning can create product capabilities that would be extremely difficult or impossible to build with manually written rules alone.

It can also fail silently, behave differently as users and data change, improve a model’s test score without improving the product, and turn a seemingly simple feature into an open-ended research project.

ML is not just another backend service. Unlike ordinary software, you cannot fully define its behavior in advance and expect it to remain stable. It depends on data, requires continuous testing and monitoring, and almost every attempt to improve it involves uncertainty and experimentation.

In this talk, we will explore:
- when ML is the right tool, and when a simpler solution is better
- how to plan effort, timelines, and expected impact when the outcome is uncertain
- what data, infrastructure, testing, monitoring, and clear ownership ML products require
- why better model metrics do not always lead to better product outcomes

The goal is not to scare product teams away from ML, but to make its real costs, risks, and requirements visible before committing to it.

And one more friendly reminder
📍 New venue: We’ll be hosted by Deutsche Telekom this time, so make sure you head to the right location!