Sense Space AI

Our research objective was to understand how AI could enhance the design research process while exploring the divide between technical capabilities and human-centered methodologies. We interviewed over 50 design practitioners across industries, focusing on design research workflows and data collection and analysis pain points. These 60-90 minute sessions explored current research processes, challenges in sensemaking, and attitudes toward AI integration. Current sensemaking processes remain labor-intensive, with designers spending approximately 40% of project time on pattern recognition that AI could automate.

We supplemented qualitative insights with computational analysis of 200+ design research artifacts, identifying patterns in how information transforms from raw data to actionable insights. Collaborative workshops with 18 multidisciplinary teams revealed communication gaps leading to implementation failures. Working prototypes of AI-assisted research tools were tested with 25 designers, providing feedback on how AI might augment human creativity in the design research process. By mapping this landscape and understanding designers’ approaches to AI integration, we established a foundation for investigating how AI might transform the design research process.

The research revealed that designers value tools that augment their intuitive and creative processes rather than replace them. Design teams spend about 22 hours weekly on unproductive sensemaking and lose over 20 hours per project to manual analysis tasks, indicating a disconnect between data sensemaking and design intuition. This gap often results in technically sound but unusable solutions or desirable yet unfeasible ones. Our findings emphasized the need for tools that eliminate time-consuming manual work, such as generating instant interview transcripts and using them to generate insights through specific design methodologies. Discussions with researchers from institutions like Parsons School of Design and MIT highlighted the importance of utilizing design methodologies as distinct elements in grounding AI tools within design research. Participants agree that a tool embedded with specific design methodologies for researchers in gathering and sensemaking data will improve their likelihood of using it through design research tasks, provided they can validate it. We discussed employing AI to use system design tools and methods like Powers of 10, the transition design approach, and methods to interact with data transcripts. This validates the growing importance of system thinking in design research and the demand for tools supporting systematic data gathering and mapping. These insights created conditions to develop a taxonomy documenting AI capabilities across design research methodologies and web applications to help designers explore these capabilities to enhance their processes.

The prototyping phase engaged 12 design researchers, 5 product managers, and five project managers from leading innovation firms like Meta, MIT, and Parsons School of Design in iterative testing sessions, shaping our development approach. Three essential components emerged as foundational for an MVP: automated interview transcription, emotional intelligence analysis, seamless integration with existing design tools, and the Validate feature, which gives researchers traceability links to where the data pipeline materialized specific insights.

Researchers unexpectedly highlighted the need for “Psycho-Portraits” – AI-generated participant profiles capturing emotional patterns and communication styles during interviews. One researcher noted, “Understanding emotional context is often more valuable than the literal transcript.” This led to the insight of developing an emotion recognition system analyzing vocal tone and facial expressions, providing deeper participant understanding. Testing revealed that 87% of researchers prioritized workflow integration over standalone functionality, such as integration with Miro and Mural, 79% valued real-time analysis during interviews, and 92% needed flexible export options for team collaboration.