Goodnotes shipped a classroom AI product by doing something most AI teams skip: sustained fieldwork. Designers Mubarak Marafa and Wendy Liao ran in-person classroom observations, documented real teacher workarounds, and built co-creation sessions with educators before writing a line of product spec. The result was Goodnotes Classroom, a 0-to-1 AI feature built on human research methods rather than assumptions.
The core argument here is structural, not motivational. Treating AI as a new design material means the discovery process has to change. When the output space is generative and unpredictable, fast prototypes tested with actual users replace static wireframes. The talk gets specific about how co-creation sessions with teachers directly shaped product decisions, which is the part worth watching closely.
This is a Config 2025 talk from Figma's conference circuit. If you build AI products and your research process is still desktop-based desk research, this is a direct challenge to that workflow. The detail on how classroom observations translated into specific feature constraints is what makes the full talk worth the time.
[WATCH ON YOUTUBE →]