AI-Based Key Product Characteristics
Influencing prompting logic & early statge strategy in an AI-based project via quick prototyping · 2025 · Allegro
AI
UX
Project goal:
Support key user tasks on the product list and improve the experience for a specific type of visit (which type - NDA).
My role & process:
I supported this project through its define, ideate, and prototype phases.
• Shifted focus from front-end delivery to improving data quality as the foundation of the experience by quickly initiating the research process. I prepared and tested prototypes built on four real data sources, combining usability testing with in-depth interviews.
• Brought UX–AI design principles and references (such as the Google PAIR framework) into our discussions, helping shape how we framed the research outcomes, which - on the other end, influenced prompting logic.
• Including the front-end stream of the project, I iterated this project during usability 3 times within ca 1,5 month - preparing rapid prototypes every time, gathering stakeholders, building shared understanding and finally shortening decision time.
Project delivered as part of my full-time role at Allegro (internal product work).
What the project gained:
• A shared shift of focus from surface design to underlying data quality.
• Input for reprioritising project phases, ensuring data improvements preceded interface work.
• Actionable design and prompt principles that informed both system logic and model prompting.
• Early product decisions — influenced by UX input — already show measurable improvement in preliminary metrics.
What I have learned:
• Never lose hope that UX can — and absolutely should — influence backend logic.
