Sebastiano Druda

Papers & exploration

Research

Exploring how artificial intelligence, human creativity, and evidence-driven experimentation can help us understand real problems before building solutions.

01

EASE 2026 · ACM · Glasgow, United Kingdom

Supporting Problem Exploration in Early-Stage Innovation Projects through Human-AI Collaboration

Daniel Planötscher, Sebastiano Druda, Xiaofeng Wang

Helping innovation teams understand the right problem before building the solution.

This research explores how human-AI collaboration can support innovation teams during the earliest stages of product development, when understanding real customer needs is often more challenging than building a solution. Through field observations of novice innovation teams, we identified recurring difficulties in problem exploration and investigated the limitations of existing AI-assisted approaches. Building on these findings, we proposed an initial design concept that uses storytelling and human-AI interaction to encourage deeper customer understanding and more evidence-driven problem exploration.

Read paperClose readerPDF · 6 pages

Author-supplied anonymized manuscript

Page — / —Fit width

Open PDF or download it if the reader is unavailable.

Text selection is available where the document supports it. The original PDF remains available above.

02

VibeX 2026 · EASE 2026 · ACM

Vibe Exploration: Human-AI Collaboration for Problem-Space Exploration in Software Startups

Daniel Planötscher, Sebastiano Druda

Bringing the speed of AI-assisted development into the discovery of real customer problems.

In this paper, we investigate how the collaborative principles behind AI-assisted software development can be extended beyond coding to help startups explore and understand customer problems. Through a case study of an early-stage software venture, we examined how AI can accelerate information synthesis and challenge assumptions, while also introducing risks such as confirmation bias and premature conclusions. Based on these insights, we proposed Vibe Exploration, a hypothesis-driven approach that combines AI assistance with continuous user interaction and evidence-based problem discovery.

Read paperClose readerPDF · 5 pages

Author-supplied manuscript

Page — / —Fit width

Open PDF or download it if the reader is unavailable.

Text selection is available where the document supports it. The original PDF remains available above.

03

Free University of Bozen-Bolzano

How to Reach Problem-Solution Fit? AI-Powered Build-Measure-Learn Loops

Sebastiano Druda · July 2026

Exploring how AI can accelerate startup learning without losing sight of real user needs.

My bachelor's thesis investigates how artificial intelligence can help early-stage startups reach Problem-Solution Fit through faster, more effective Build-Measure-Learn cycles. Combining academic literature, industry insights, and case studies, I explored how AI can support problem identification, experimentation, and customer feedback analysis. I also examined two of my own projects, FiscoBud and Bundyclock, to understand the practical opportunities and limitations of AI-assisted validation. The research highlights a central principle: AI can accelerate learning and experimentation, but meaningful validation still depends on real users, critical thinking, and sound human judgment.

Read paperClose readerPDF · 53 pages

Bachelor thesis

Page — / —Fit width

Open PDF or download it if the reader is unavailable.

Text selection is available where the document supports it. The original PDF remains available above.