Experience-Driven Design for Generative AI Integration in a Real Estate ERP System
Malada, Ariel (2026)
2026
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
Hyväksymispäivämäärä
2026-05-07
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202605054953
https://urn.fi/URN:NBN:fi:tuni-202605054953
Tiivistelmä
Generative AI (GenAI) adoption is growing rapidly in the recent years. A survey representative in the US found 10% adoption increase in just a year from August 2024 to August 2025. However, UX focused research on GenAI is still limited for specific case enterprises system like real estate ERP system.
This study developed generative AI feature for the OviPro real estate ERP system using Experience Driven Design (EDD) and combined the approach with Weisz et al.'s (2024) design principles for generative AI applications and applied it across the three study phases. The first phase is user research by using parallel mixed questionnaire method combining User Experience Questionnaire Short version (UEQ-S) and open-ended questions. The questionnaire was distributed to Finnish real estate agents (n=68) to understand their current AI tool experience in their daily workflow. The questionnaire analysis using UEQ-S result and thematic analysis for the open-ended responses shows three identified negative experiences: inefficient, unattractive, and fearful. Three experience goals formulated based on the negative experiences are: Trust and Reliability, Control, and Efficiency. In the second phase, a generative property description feature was designed and developed, with the experience goals and Weisz et al.'s (2024) design principles to guide the design decisions through iterative process. The third phase evaluates the feature through a contextually embedded mixed-methods questionnaire (n=21). UEQ-S results showed improvement in both pragmatic quality (from 1.11 to 1.42) and hedonic quality (from 0.89 to 0.92) compared to the first phase. The qualitative findings using thematic analysis validate the Control and Efficiency experience goals, while Trust and Reliability was partially achieved from the interface elements and design mechanisms but affected by the language model output quality.
The study presents that combination of EDD and generative AI design principles offers a working framework for developing AI features in enterprise systems. Additionally, the findings highlight the importance of separating content level trust and design level trust in developing generative AI features as content level trust depends on the quality of the language model output that beyond designer’s control.
This study developed generative AI feature for the OviPro real estate ERP system using Experience Driven Design (EDD) and combined the approach with Weisz et al.'s (2024) design principles for generative AI applications and applied it across the three study phases. The first phase is user research by using parallel mixed questionnaire method combining User Experience Questionnaire Short version (UEQ-S) and open-ended questions. The questionnaire was distributed to Finnish real estate agents (n=68) to understand their current AI tool experience in their daily workflow. The questionnaire analysis using UEQ-S result and thematic analysis for the open-ended responses shows three identified negative experiences: inefficient, unattractive, and fearful. Three experience goals formulated based on the negative experiences are: Trust and Reliability, Control, and Efficiency. In the second phase, a generative property description feature was designed and developed, with the experience goals and Weisz et al.'s (2024) design principles to guide the design decisions through iterative process. The third phase evaluates the feature through a contextually embedded mixed-methods questionnaire (n=21). UEQ-S results showed improvement in both pragmatic quality (from 1.11 to 1.42) and hedonic quality (from 0.89 to 0.92) compared to the first phase. The qualitative findings using thematic analysis validate the Control and Efficiency experience goals, while Trust and Reliability was partially achieved from the interface elements and design mechanisms but affected by the language model output quality.
The study presents that combination of EDD and generative AI design principles offers a working framework for developing AI features in enterprise systems. Additionally, the findings highlight the importance of separating content level trust and design level trust in developing generative AI features as content level trust depends on the quality of the language model output that beyond designer’s control.