Before this class, I’d gone through the design process in CS147. I had experience with needfinding, rapid prototyping, and usability testing, some overall knowledge. What I didn’t expect was how much new territory there still was to cover, and how differently it would feel to move through a design process when the problem space was something I cared about more than in my last class.
One thing that genuinely excited me in this class was going deeper on branding and visual identity. That’s a part of design I’d been drawn to but never formally studied, and even though the module in class was short, I felt it was very valuable.
Mental models like the iceberg model gave me a new way to think about what’s actually driving user behavior beneath the surface. The weekly blog format worked surprisingly well for me. Submitting work incrementally kept the project from collapsing into a deadline crunch and made it easier to actually reflect on what we were building as we went, rather than just shipping and moving on.
The thing I’ll probably carry longest from this class is the approach to using AI in the design process — yes as a shortcut, but also as a creative collaborator. This was the first class where we jumped straight into AI-assisted prototyping, and it felt immediately relevant to the kind of work I can do now outside of class. Learning to design with AI — using specific prompts, style tiles, and mood boards to avoid the generic flatness of AI slop — changed how I think about what’s actually possible to build on your own. It made the work feel motivating in a way, because some version of the product was now available really quickly. The gap between idea and artifact is a lot smaller now.
The tension at the heart of our project is that we wanted to build something that got people off their phones. In our case this was before bed to improve sleep quality, but part of our solution required an app. That contradiction lived with us the whole quarter. The teddy bear felt like the real intervention: a physical presence in the bedroom environment, something warm and familiar that could deliver content without a screen. But we didn’t have the resources to build it out, so what we shipped was the companion app for a product that doesn’t fully exist yet. That gap between vision and prototype is still unresolved for me. I think the app is genuinely useful, but it only makes full sense when Teddy is in the room.
The behavior change content hit differently for me this quarter because I was living it in parallel. I took a writing class alongside this one and was trying to build a consistent writing practice. What actually worked was attaching it to something I already did: going to the gym in the mornings. After the gym, I’d read and write. I started carrying a notebook everywhere and a book. I got into reading again — I read 4 books this quarter! The habit stacking framework we studied in class was exactly what I was doing in my own life, and watching it work made me trust it as a design principle.
On ethics: Our nudges depend almost entirely on the teddy bear being physically present in the bedroom. That’s intentional — a physical object nudges through presence and emotional attachment rather than notifications or dark patterns. What makes it an acceptable nudge rather than a manipulative one is that users set their own goals, choose their own content, and name their own bear. The system is working toward something the user explicitly defined. The risk of manipulation is highest if the product were ever designed to exploit that attachment — making users feel anxious without Teddy, or maybe creating dependency through variable reward patterns the way social media does. We tried to design against that.
Privacy was a real design constraint. A device that lives in your bedroom and responds to your presence raises obvious concerns about ambient listening. We resolved this by deciding that Teddy would be a speaker only — no microphone, no always-on listening.
On inclusion: our persona work centered around college students, which was appropriate for the scope of the class but real people who struggle with sleep don’t fit neatly into that demographic. Students with ADHD, anxiety disorders, or chronic insomnia have meaningfully different relationships with bedtime routines, and our current design doesn’t account for that. The commitment lock feature, for instance, could increase anxiety in users for whom rigid structure is itself a stressor. That’s an edge case we didn’t fully design around.
What I think I’ll remember ten years from now is the habit stacking as something I actually used to change my own life during the quarter I was studying it.
Next time, I’d like to address the hardware gap. We knew from the beginning that the app alone wasn’t the full solution, but we let that tension sit in the background rather than making it the central design problem. I’d want to spend more time on what the minimal viable physical component could look like, even a simple prototype, so the app has something real to be a companion to.
