Ehxale
Vibe-coding a 7-day guided reflection tool
Observation
So much within the wellness + tech space focuses on maxxing and optimizing.
Instead, I wanted to experiment with something that does the opposite:
A moment to exhale, to empty the cup.
“Empty your vessel, and receive what is ready to emerge.”
Context.
For many of us, unrealized decades of chronic stress and anxiety have been accepted as a consequence we inherit to participate in the corporate working world. But without a way to process what we've accumulated, that stress surfaces as physical ailments and autoimmune conditions (from eczema flare-ups, IBS, tight chests, shoulders - you name it)
Part-documentation, part-fascination, part-just me being ever curious about new technologies, I decided to experiment with vibe-coding a simple tool based off tactics that have helped me with processing the built-up that’s been living in my nervous system.
Design ethos.
Organic and kinetic.
Acknowledging that our thoughts, memories and emotions are fluid, and non-liner.
Holds space for contemplation.
A vessel suspended in liminal space. Holds space for reflection. It does not impose, and only witnesses.
Somatic anchoring.
The method of this journal should integrate what’s on our mind with what’s held in the body.
Key Features (& what I learnt building them)
The 7-day experience arc.
As a momentary journal-er, I’ve found this practice immensely helpful in times of acute stress (e.g. rumination, overthinking, swirling thoughts), or when I’m yearning for some clarity (ie when I’m ‘in my head’ a lot). The arc is not designed to be habit-forming. Move through it once, revisit it often, or repeat the journey whenever you're ready. Each pass offers the chance to notice something different, and receive something new.
The AI-generated reflections.
Not designed to advise, but to make you feel witnessed. The voice of responses shaped by some of my favorite writers and philosophers: Alan Watts and his comfort holding contradiction without needing to resolve it. Joan Didion's precision and refusal to sentimentalize hard things, especially the sensibility running through The Year of Magical Thinking. Rick Rubin's The Creative Act, and its idea that creativity is a practice of attention, not output.
The modalities of metabolizing thoughts.
I deliberately chose a free-form style of journaling to begin the process of an ‘exhale’. Flooded thoughts are never coherent - akin to our memories. The act of writing gives them somewhere to go - onto a blank canvas, and out of our heads.
I tried to create a ‘walk and talk’ feature with the mic option (for verbal processors), but unfortunately, the existing tech I have is not suitable for mobile users. Stay tuned.
A guided meditation features in one of the days, incorporating a simple somatic visualization exercise - a 'threshold’ moment that breaks the exhale to a gradual, inhale. While building this feature, I encountered several barriers with trying to code in the right pacing of the words showing up on the page. Turns out, the best way to get this pacing right was to record my own voice and have this encoded.
A safe vessel.
Thoughts are just thoughts. Once metabolized, there is no reason for the to linger. Journal entries are permanently deleted when you complete the 7-day experience. Entries from unfinished journeys are permanently deleted after 7 days from your last activity.
Learnings I’ll be taking forward
From designing the tool.
Reject the playbook of linearity and optimization. Instead, learn from the intelligence of biology. For example: Nature doesn't create in isolation - it is an amalgamation of different datapoints, interconnected to each other. Nature also does not measure itself in speed and efficiency - instead, it respects its seasons, and thrives when in balance with its environment.
Secondly, we (aka the ones vibe coding) get to choose if the tool we create is additive - or extractive - to the human experience. Choose to be additive.
From vibe-coding.
Building your own stack is a bootcamp in understanding AI. A few things I got sharper on:
Using different models as different thinkers, like you would when pressure-testing a strategy of framework with a set of diverse thinkers. My personal preference: Claude for the deep, structural work. Gemini for fact-checking. Chatgpt for quick edits, never for code.
As a non-developer, the ability to own a real technical stack, from building features to making security decisions is pretty impressive for a (very) viable proof-of-concept.
The product development process is familiar for any innovator, just with different components.