Magnets

Magnify the moments that make you, you.

Team: Sarah Mou, Andrew Liu, Julian Jung, Joanna Ye
School: University of California, Berkeley
Event: FigBuild Hackathon, March 2026

Inspiration

The four of us are students at UC Berkeley. Two of us are currently sophmores and the other two remember their first years well enough to know how much of it is hard to account for afterward.

The big events you always remember. What is harder to hold onto is everything else: the specific bench you kept going back to between classes, the conversation that went until 3am, the first time a city you moved to started feeling like yours. You know those moments happened and you know they mattered, but you cannot point to them or go back to them in any concrete way.

When we sat down with the prompt, each of us had a slightly different way of describing the same gap. It took us longer than expected to land on language that actually named it, which turned out to be the most useful part of the whole ideation process. We were all talking about the felt sense that a moment belongs to your identity, that something about it is distinctly yours, and that you have no way to capture that sense before it fades.

We eventually called it identity resonance and once we had a name for it the rest of the project followed from there.

"I know I changed freshman year. I just can't prove it." — Joanna, 20, Berkeley CA

What We Built

Magnets is a mixed-reality identity capture tool. It passively detects moments of identity resonance throughout your day using biometric wearables, then crystallizes those moments into illustrated sticker artifacts called magnets. Users collect magnets and arrange them on a virtual fridge, which becomes a growing visual record of who they are becoming over time.

The product consists of three surfaces working together.

The Wearables Layer

A pair of Meta Ray-Ban glasses and a wrist-worn biometric sensor operate in the background throughout the user's day. They track heart rate variability, skin conductance, dwell behavior, and ambient context to detect physiological signatures of resonant moments. No active input is required from the user.

The Mobile App

When a resonance threshold is crossed, the moment is queued for review. Users open the app and see a maximum of three queued moments per day so the experience never becomes a feed to consume. They review each one, decide whether to stamp it into a magnet, and the stamping process renders the photo in an illustrated sticker style with metadata automatically attached: location, date, time, and source device.

The Fridge

The fridge is a spatially organized, time-bounded collection of magnets named for a specific period of life. Users arrange magnets on the fridge themselves the way you would arrange photos and stickers on a real refrigerator. Clusters emerge naturally over time and the fridge starts telling a story the user did not know they were living.

The New Sense: Identity Resonance

Emerging science has identified anywhere from 22 to over 33 distinct human senses, most of which operate below conscious awareness. One of them has no tool yet.

Identity resonance is the felt sense that a moment, place, person, or object belongs to your identity. It sits closer to interoception than to the classical five senses. Researchers studying the relationship between interoception and self-concept have found that internal physiological signals contribute directly to the stability of identity over time, and that the core features of the self-concept are those that correlate most with inner bodily states.

Eugene Gendlin, the psychologist who pioneered the concept of the felt sense, found that people who could connect to their bodily experience were far more likely to undergo lasting personal growth. The felt sense is the language of the nervous system. It has been running in the background of every meaningful moment you have ever had, and it has never produced a record.

To quantify when a moment crosses the resonance threshold, we modeled detection as a weighted combination of biometric signals sampled at time $t$:

$$R(t) = w_1 \cdot \Delta \text{HRV}(t) + w_2 \cdot \text{EDA}(t) + w_3 \cdot D(t)$$

Where $\Delta \text{HRV}(t)$ is the deviation in heart rate variability from the user's rolling baseline, $\text{EDA}(t)$ is the electrodermal activity reading normalized to the user's personal range, and $D(t)$ is a dwell score capturing how long the user's attention remained on a particular context. The weights $w_1, w_2, w_3$ are initialized equally and adjusted over time through user feedback as the system learns which signal combinations are most predictive for that individual. A moment is queued when $R(t)$ exceeds a threshold $\tau$, where $\tau$ is set at the 85th percentile of the user's personal resonance distribution:

$$\tau = F^{-1}(0.85)$$

This means the system surfaces only the top 15% of moments by resonance score, keeping the daily queue small and meaningful rather than flooding the user with everything that registered above a flat cutoff.

Who It Is For

Magnets is designed for college students aged 18 to 24, specifically those in their first year. There are currently 19 million undergraduate students enrolled in the United States and 57% of them report feeling lonely, with 70% saying they have struggled with their mental health since starting college. Most of that happens in the first year, when the social and emotional infrastructure of a person's life is being rebuilt from scratch.

The problem is not that these students lack self-awareness. The problem is that the moments doing the most significant work in shaping their identity are the ones that leave the least trace after they happen.

Our primary user is already identity-aware and visually literate, already in the habit of saving things that feel like them, and looking for a way to build a record of their life that is for themselves rather than for other people to see.

Wellness Goal

Magnets addresses emotional and social wellbeing. The goal is to help users who are in transitional periods of their life develop a clearer and more grounded sense of who they are, particularly during the first year of college when their social environment, daily routines, and sense of belonging are all being established at the same time.

The behavioral shift we are supporting is a gradual move away from the kind of passive, externally-oriented identity consumption that characterizes most social media use toward something more reflective and internally grounded. Rather than building a picture of yourself through what you post or how others respond to you, Magnets builds that picture from the physiological record of what actually resonated with you over time.

We were deliberate about keeping the product free of optimization mechanics. There are no scores or recommendations telling users what to do differently. The product's value is in showing users what is already there.

Managing the New Information

We thought carefully about the risk of surfacing a new kind of perception only to overwhelm the person receiving it. The information design of the app reflects that concern at every level.

The wearables capture data throughout the day without any active input from the user. When the app has moments to surface, it shows a maximum of three per day, which keeps the experience manageable rather than like something to keep up with. The user reviews each queued moment and decides whether to stamp it into a magnet. Nothing lands on the fridge without that deliberate choice. Each fridge is also bounded by a named time period, so the collection stays oriented around a specific chapter of someone's life rather than growing into an undifferentiated archive.

Pattern recognition and any form of insight surfacing are entirely opt-in. The fridge itself is spatially arranged by the user, so the clusters and groupings that emerge are the ones the user builds rather than ones an algorithm decided to highlight.

Safeguards

Privacy and Data Ownership

All data is stored on-device by default and never sold, shared, or used to train models of any kind. The app does not record audio and does not perform facial recognition on other people. No location data about other people is ever stored. When another person appears in a queued moment, their features are automatically blurred before the user even sees the preview.

Painful Moments

A moment can register as resonant and still be something the user does not want to keep. If a queued moment is flagged as painful, the app does not stamp it automatically. The user can archive it privately, place it behind a soft blur on the fridge, or delete it permanently. Deletion is immediate and irreversible with no cloud backup. The app never resurfaces past moments unprompted and has no on-this-day notifications.

Other People's Consent

Any other person appearing in a magnet can only be named with their explicit approval granted inside the app. No identifiable information about other people is stored or shared under any circumstances.

No Optimization

There are no scores, streaks, or metrics of any kind in the product. We designed specifically against the scenario where users start treating their resonance data as a performance to optimize, because that would turn the tool into exactly the kind of externally-oriented pressure the product is supposed to offer an alternative to.

How We Built It

We built Magnets entirely in Figma over the course of the hackathon. The full prototype was designed in Figma Design and connected using Figma's native prototyping tools, covering five core flows across the product: onboarding, fridge creation, the passive moment detection experience, the stamping flow, and the fridge view where users arrange their magnets.

We used Claude Opus 4.6 via Figma Make to assist with structural decisions during the design process, particularly when thinking through information hierarchy and the logic of how a moment moves from the wearable layer into the app. For the magnet visual style, we used Midjourney to develop an illustrated aesthetic that made each magnet feel handmade and physical, keeping the style consistent across subjects as different as a red Ferrari, a Shiba Inu, and a bowl of ramen.

None of us came into this hackathon with any significant Figma prototyping experience. We learned those tools during the event, and getting to a working prototype we were satisfied with by the end is something the team is genuinely proud of.

Challenges

The hardest part of the process was ideation, and it took longer than we expected. Each of us came in with a different version of the same general feeling: that something important was happening in daily life that no existing tool was capturing. But every time we tried to describe it, we ended up describing something that already existed. We kept arriving at products that sounded like mood trackers or memory apps or journaling tools with a slightly different interface.

What eventually broke the logjam was slowing down and asking a more specific question: not what was underserved in the existing landscape, but what produced no signal at all. That question led us to identity resonance and to interoception as the scientific foundation for the concept.

The prompt also redirected our thinking in a way we did not anticipate. Our instinct as students is to think about the future in terms of our own lifetimes, what we will be building or using in the next ten or twenty years. The challenge asked us to design for generations we will not be part of. That shift in perspective changed how ambitiously we framed the problem and made us more willing to commit to an idea that was genuinely speculative rather than something we kept hedging toward feasibility.

What We Learned

The most useful design decision we made was committing early to a single named user and following her through every flow and every edge case. Joanna is not a persona we used for a slide and then set aside. She is someone with a specific year, a specific campus, and a specific bench she kept going back to, and we ran every decision through the question of whether it would actually serve her. That constraint made the product more coherent than it would have been if we had been designing for an abstract user type.

We also found that saying no to features was where most of the real design work happened. The principle we kept returning to was that the product should show you what is already there rather than tell you what to do about it. That ruled out streaks, scores, push notifications, social sharing, and a feed model entirely. Each of those felt like a loss in the moment but the product is much more focused for the absence of them.

What's Next

The most immediate next step for the product is hardware. The resonance signal gets richer and more precise as the underlying biometric data gets more comprehensive, and wearables are becoming significantly more capable each year. A version of Magnets built on denser sensor data would surface moments with more accuracy and produce fewer false positives in the detection layer.

Two features we scoped but did not build are shared fridges and multi-period navigation. A shared fridge would let two people who keep showing up in each other's resonant moments build a joint record of a relationship, with explicit mutual consent required from both people through the app. Multi-period navigation would let a user move between all of their named fridges in a single view, so the full arc of multiple chapters sits alongside each other rather than in separate collections.

The longer-horizon question the prompt asked us to sit with is whether the core concept scales beyond the freshman year use case. The underlying observation is that the body has always been registering which moments matter, and no tool has ever been built to read that signal and make it persistent. That is a problem that applies at every stage of life, and the fridge model is one way of structuring an answer to it.

The Team

Name	Major	Class
Sarah Mou	Cognitive Science and Design	2027
Andrew Liu	Computer Science and Data Science	2028
Julian Jung	Business and Global Management	2028
Joanna Ye	Cognitive Science, Data Science, and Design	2026