A Comparative Analysis for GenAI-Enabled Learning Systems for Urban Portfolios in ECIS

The UNDP City Experiment Fund (CEF) supports municipalities across Eastern Europe and Central Asia as they navigate a “Twin Transition” toward green and digital development. These cities rely on urban portfolios to test interventions, yet learning remains fragmented, slow, and difficult to reuse. Traditional methods such as workshops, reports, and manual sensemaking often fail to capture timely, actionable insights.

This Capstone project evaluated whether Generative AI (GenAI), specifically the PIA (Portfolio Intelligence Assistant) tool, can improve learning systems within CEF. The approach combines three components: system testing using a structured evaluation rubric, survey analysis of user experience, and in-depth stakeholder interviews to understand real-world adoption and constraints. These methods allowed for a comparison of AI-enabled learning against traditional human-led processes.

The findings showed that PIA performs well in structured learning capture and intent recognition but faces challenges in validation logic, user effort, and insight generation. Interviews further revealed that PIA is most effective as a complement — not a replacement — to existing practices, with human verification remaining essential.

Based on these insights, the team proposed a phased roadmap. Phase 0 focused on improving usability, reducing friction, and standardizing “good learning.” Phase 1 emphasized adoption, data quality, and demonstrating value through summaries and pattern detection. Phase 2 envisioned PIA as an embedded workplace assistant supporting daily workflows and knowledge retrieval. The final report provided actionable recommendations on where GenAI could enhance urban learning systems and where organizational and human-centered approaches remain critical for effective governance.