Still Feed · 虚室生白

A design research project on tangible LLM interfaces and artificial mindfulness · Prototype · in progress

“虚室生白,吉祥止止。
An empty mind breeds pure enlightenment; total stillness attracts boundless blessings.”
— Zhuangzi

Less, on purpose

I keep wondering whether AI is making life easier or simply making it busier.

Most AI interfaces give us more: more text, more answers, more notifications. They connect to our email, calendars, notes and files. Now several agents can work across parallel threads, all behind the same little sparkle icon that seems to have appeared in every piece of software. Starting something has become almost effortless. Paying attention to it hasn't.

Are we becoming more productive, or just starting more half-formed tasks? How much of what we ask AI to produce do we actually read, think about or use?

These questions led me to a different kind of interface. Could AI stay in the background? Could it be tangible, screenless and a pleasure to live with? What if it could read the weather, my calendar, signals from my body and my intentions, then give back almost nothing?

Just enough to set the tone of a day. Quiet enough to ignore. Beautiful enough to leave on the wall.

The thing I'm making

Still Feed is an ambient AI frame that reads my day and writes it back in ink. Sometimes it offers a few words. Sometimes it draws a timeline. Sometimes it makes a circle that invites me to breathe.

A pen plotter sits behind a hanging scroll and writes on the back of the paper. From the front, the ink slowly seeps through, and the strokes seem to appear on their own. I'm using one of our oldest tools, pen on paper, to explore how we might live with AI more quietly, in a calm and subtle way. And how can AI treats what gets shown with great care, and to make room for a moment of stillness.

I’d like to call this artificial mindfulness: an interface that practises restraint on my behalf. It reads more than it writes, takes its time and is allowed to say nothing at all.

The name 虚室生白 comes from Zhuangzi. The image of an empty room filling with light feels close to what I'm trying to make: something that leaves enough space for a person to think and feel.


Interfaces for artificial mindfulness

I want to explore this at home environment, where most of life happens. It holds our energy, our emotions, our routines and our time with each other. As homes today fill with connected devices, they also fill with screens, speakers and things asking for attention. It is becoming where an ambient AI interface is needed most. Therefore, I'm interested in how information can live at the edge of attention in domestic environment: present in the room, readable at a glance and easy to leave alone.

For Still Feed, that means three things as an artificial mindfulness device:

  • Care in what it shows. It chooses a few things that matter from everything it knows. Silence counts as an answer.

  • Gentleness in how it appears. It waits to be glanced at. Paper, ink and physical movement give information a presence without a glowing screen or a notification.

  • A connection to the person. A calendar holds obligations. A person also has intentions, moods, a body and things they want to make time for. The interface needs to understand that fuller picture, and let it change.

These ideas take shape in four sheets: the Dashboard, the Line, the Almanac and Breathe. Each starts with a different question.

The Dashboard:
how should I start?

The dashboard is the default sheet. A scheduled LLM task runs at set times of the day. It pulls in everything the day holds, the weather, my calendar and my WHOOP data, and decides what the sheet should say.

It also draws on my personal Obsidian vault, where I keep notes, plans and reflections, and on occasional long-form conversations with an AI about what I'm working towards and what I want more time for.

The model weighs the day against that picture and keeps at most three priorities. I want it to put what I care about ahead of what merely asks for me. The rest stays off the sheet. The aim is to help me begin from my own intentions before the day's demands take over.

At the top, it offers a hint in four Chinese characters. On a clear Tuesday with nothing booked, that might be 静以養神: “be still to nourish the spirit.”

Why four characters? Chinese has a long tradition of holding a lot in a small space: chengyu idioms, New Year banners over doorways, the names of the twenty-four solar terms. Calligraphy adds another layer of poetic quality that suits stillness. The ink, the pauses between strokes and the empty paper around them all contribute to the meaning.

Together, the limit asks the AI to find the gist of a day. It also leaves me room to decide what the words mean for me.

The Line:
when am I free?

Some days have only a few fixed appointments: a call at 10:30, perhaps, with open time around it. The Line draws the day as a ruler, from six in the morning to midnight. Small marks show the hours; larger, labelled marks appear every three hours. Each commitment becomes a heavy dash. Everything else stays open paper.

What I want to see is how much of the day is still mine.

One idea I want to test is whether the most useful message might be 无事: “nothing pressing.” We often reach for our phones because we're afraid of missing something. Could a trusted “nothing needs you” on the wall make it easier to leave the phone in another room?

The Almanac:
what does my body allow?

The Almanac borrows its form from the Chinese Huang Li, which names what a day is suited to (宜) and what to avoid (忌). Here, those suggestions respond to signals from my body.

My WHOOP strap estimates recovery using heart-rate variability, resting heart rate and sleep. The sheet draws that score as a partial ring and uses recovery, sleep and strain data to suggest what kind of day to make room for.

When recovery is especially low or high, I want that information to come before the calendar. An open afternoon can look the same in a schedule while feeling very different in the body. This sheet asks the frame to notice that difference.

Breathe:
can I stop for a minute?

Breathe is an invitation to pause. The frame feeds a fresh sheet, draws an ensō, a Zen circle traditionally made in a single stroke, then keeps moving just above the paper at the pace of a breath.

There is very little to read. Just a circle and a slow movement to follow. I can join in or let it happen in the background.

Eventually, I'd like this sheet to respond when stress rises during the day. WHOOP tracks stress in its app, but doesn't currently expose that score through its API. For now, I use a very low morning recovery score as an indirect cue, or trigger the sheet from my phone when I notice I'm overwhelmed.

This is part of what I want to explore: whether something slow happening in the room can make a moment of stillness easier to enter, without turning it into another task.

How it's made

The frame is a CoreXY pen plotter, driven by an Arduino with a CNC shield and rebuilt to stand upright. A roll of Xuan paper (宣纸) hangs in front of the drawing area. Mirrored G-code lets the pen write on the back, keeping the mechanism out of sight while the ink seeps through.

The pace matters to me. One character takes about forty-five seconds; four take roughly as long as a pot of tea takes to steep. That slowness is part of the experience I'm designing.

The software runs locally through a controller interface I co-created with Codex. One of the harder problems is cursive Chinese (草书). A font describes the shape of a character, but doesn't tell a pen how to draw it. The backend renders each glyph as an image, traces its centreline and arranges the paths into strokes, using the regular character's stroke order as a guide. Those strokes are laid out, mirrored and sent to the plotter as G-code.

Living with it

Still Feed is still a design probe: something to live with, observe and learn from. Over the coming months, I want to keep it on my wall, try all four sheets and pay attention to what changes.

Do I reach for my phone less in the morning? Does “nothing pressing” become something I can trust? And when the frame gets it wrong, what does a wrong answer feel like when it's written in ink and can't be taken back?

The project brings together several of my research interests. Calm technology asks how information can move between the centre and the edge of attention. Slow technology makes room for reflection and rest. Tangible and ambient interfaces bring information into the physical spaces we inhabit. Still Feed is also an expression of my growing wish for a more mindful, screenless life.

LLMs give these questions a new weight. As a system gains access to more of my day, it has more to interpret, and more to leave out. I want to understand how it should make those choices, and what form the small amount it shares should take, how much we as human should trust such a system and the delegations.

Still Feed is my starting point for that exploration. I'm trying to find out what artificial mindfulness could mean as a design goal, and whether a machine can help make room in a day by knowing when it has said enough.

Next
Next

Rewind