Synthetic data is safe. It is also often too perfect. Real-world data contains messy language, missing information, strange formats and unexpected patterns that expose where our solutions fail.

So how do we use real-world data responsibly? How do we keep the story but leave the identity?

Behind every record is a real person, and using data beyond its original intent carries responsibility. Yet without those real stories, our solutions can become a fairy tale.

For engineers and architects, this talk explores how to use real-world data responsibly. We’ll look at why redacting obvious personal information is not enough, how context can reveal identity, and how replacing sensitive data can change the meaning of the original story.

You’ll leave with a practical way to think about what data you need, what to redact, what to retain, and how to validate the result.