Gustav Soderstrom

Co-President, CPO and CTO at Spotify

Co-President, Chief Product Officer and Chief Technology Officer at Spotify, responsible for global product and technology strategy, who has been at the company for over 14 years growing from leading mobile to overseeing all product, design, data, and engineering.

The Visionary The Operator

Dimension Profile

Strategic Vision 95%

Execution & Craft 70%

Data & Experimentation 60%

Growth & Distribution 45%

Team & Leadership 60%

User Empathy & Research 75%

Key Themes

curation to recommendation to generation era fault-tolerant user interfaces for AI product magic tricks for virality AI DJ as generative product innovation internal podcasting for culture building generative AI impact on music industry

Episode Summary

Gustav Soderstrom shares how Spotify navigates the shift from curation to recommendation to generation, introducing principles like fault-tolerant user interfaces that match the actual performance of AI models and the importance of product magic tricks for virality. He walks through the development of the AI DJ as Spotify's first truly generative product, explains why great AI products should do as little as possible and get out of the way, and reflects on using an internal podcast to build culture and make senior leadership more approachable across a massive organization.

Leadership Principles

→ The internet went from curation to recommendation to generation — each era requires rethinking your entire product and sometimes your business model
→ Design fault-tolerant user interfaces that match the actual performance of your AI — if your model hits one in five, show five things on screen, not one big play button
→ All truly great products pull some magic trick — iterating a product to the point where it feels like magic the first time correlates with going viral

Notable Quotes

"The internet started with curation, then switched from curation to recommendation, and I think what we're entering now is generation. I suspect it will be as big of a shift that you will eventually have to rethink your products."
— On the three eras of internet products and why generation changes everything

"If your machine learning hits one in five, you need a UI that shows five things on screen. A single big play button requires literally 100% or zero prediction error, and that's never the case."
— On the principle of fault-tolerant user interfaces for AI products

"The principle for the AI DJ was literally to do as little as possible and get out of the way. It's not telling you what the weather is. It is trying to get you to the music."
— On restraint as a design principle for generative AI features

Want to know how you compare?

Take the Assessment