Why it mattered

Returning users often land on generic pages that do not reflect what they are actually trying to do. I wanted to explore how AI could help create a more tailored experience.

Key decisions

• Predicted intent based on previous user activity even when not logged-in
• Identified user profiles (guitarists, drummers, producers etc.)
• Used an LLM to generate a tailored welcome message
• Used AI APIs like “Google’s Recommendation” to build product bundles

Outcome

As an exploratory concept, success was assessed through user testing, where 70% of participants agreed that the recommendations matched their interests. The other 30% felt indifferent or did not find it useful.

Personal reflections

This exploration helped me understand how different AI services can work together to create personalised experiences, while also supporting the goal of increasing revenue per visitor.

Why it mattered

My son Liam (8 months old at the time of writing) is growing up in a bilingual English–Italian family. I wanted to support early language development by helping him associate objects, people and sounds with pronunciation during the critical 6–18 months.

Key decisions

• Focused primarily on visual and audio association
• Used big tappable areas to maximise usability
• Used native pronunciation in both English and Italian
• Included real-world sounds for animals and music instruments

Outcome

Although it is still too early for Liam to pronounce words, after a week of use he consistently reacted more to certain musical instruments and animal sounds, suggesting early recognition and preference.

Personal reflections

Designing for children requires extreme simplicity. This is a parent supervised learning tool rather than a passive entertainment app, reinforcing how clarity and focus can create calm, effective educational experiences.

Why it mattered

“Bedroom producers” struggled to achieve polished, professional sounding guitar tones without expert sound engineering knowledge.

This created an opportunity to simplify a complex mixing process and help beginners achieve better results, faster.

Key decisions

• Automatic detection of guitar type to improve sound accuracy
• A single, intuitive control instead of multiple technical parameters
• Focused on “bedroom producers”, not professional engineers

Outcome

This became the most popular product across 2019 and 2020 combined, generating the highest overall income in my product catalogue.

Personal reflections

I strongly believe that tools like this should simplify the music production workflow, that’s why a single control was the key of its success.