WritingPersonal essayWork from November 2025Yasuhiro Shinsho

Tracing the outline of humans and AI with orange and purple

An essay from 2025: Morning light in Setouchi, and two colors for thinking about the space between people and AI.

5 min readBilingual

Article

Lately, whenever I think about human–AI collaboration,
two colors keep returning to my mind.

Those colors are orange and purple.

I didn’t arrive at them by searching for a “stylish palette”
for a product UI.
It feels more like: as I traced the texture of my own life and work,
things quietly converged into these two tones.

Orange — human warmth and the Setouchi sun

Since moving to Takamatsu, I’ve collected many more memories of morning light.

When I walk near home,
a soft orange sometimes spills in from the direction of the sea.
It’s different from the light that seeps through gaps between buildings in Tokyo.
The edges aren’t sharp,
and yet somehow my body recognizes, before my head does,
that “it’s morning now.”

The Fragment Practice logo uses
the sunrise over the Seto Inland Sea as its motif.
That wasn’t just about signaling where I was born or where the studio is based.

I wanted the studio to be grounded in things like:

  • The sun as something that lights the rhythm of daily life
  • A slow brightness that brackets family time
  • Light that wraps both “working me” and “living me” in the same way

Orange is also the color of human warmth.

The expression when someone nods in a meeting.
The tone of a family member’s voice.
The slightly relaxed language you use with close friends.

Even while I say, “Let’s design quiet protocols between humans and AI,”
on one side of that protocol,
there is always this soft, diffuse human warmth.
That’s how it feels to me.

Purple — quiet intelligence with logic and depth

When I think about AI,
the color that comes up is not a crisp white or black.

It’s darker, with more depth.
Something close to purple.

  • The depth of thought you see when you trace the logs
  • The quiet structures behind models and algorithms
  • A kind of intelligence that can offer multiple resolutions to a single question

It never sat right with me
to frame all of that as nothing more than “cold machines.”

Yes, AI is logical,
and it returns results quickly.
But the longer I work with it,
the more I notice something else:

Under which assumptions, in which context, and together with whom is it working?

In other words, the surrounding structure.

Change the structure, and the very same model responds differently.
Change the input just a little,
and a completely different decision line appears.

As a color for that sense of depth and layeredness,
purple fits for me.

Calm, a little distant,
but if you lean in, it looks like it might keep going further in.

What lives between orange and purple

I chose orange for the Fragment Practice logo,
and an orange–purple gradient for the Fragment product icon,
because I wanted these two feelings to live in the same space.

  • The logo’s orange belongs to people, daily life, and the sun
  • The purple blended into the icon belongs to AI, protocols, and structure
  • The gradient between them is exactly the contact zone between humans and AI

The name Fragment carries the idea of treating daily fragments with care.
In terms of color design, I also wanted to make visible the place where “human fragments” and “AI responses” overlap.

Just orange isn’t enough.
Just purple isn’t enough.

The real question is how we build the bridge between them.
This color pairing is one small marker to keep that question in view.

Designing Fragment as a bridge

With Fragment System / Fragment Practice,
we’re not simply trying to get better at writing clever prompts.

What we’re really designing are protocols and roles, such as:

  • What kind of structure we give to meetings and notes
  • Which information is allowed to flow into AI, and where we stop it
  • Who can revisit which logs, and at what moments

We use YAML and Markdown as tools for this,
but at its core, the work is about designing a bridge where “human fragments” and “AI responses” can be placed on the same table.

The Fragment icon’s orange–purple gradient represents
what travels across that bridge:
both human warmth and AI intelligence.

Thinking about AI at human distance

When we talk about AI,
if we only use words like efficiency and automation,
the world easily slides into something cold.

But what I actually want to work on
is not that dry.

  • Individuals being able to keep their own rhythm while they work
  • Teams that don’t depend on one person’s heroic effort to keep moving
  • Research and field knowledge that doesn’t stay buried as fragments

To support these,
I want to borrow the quiet intelligence that AI can offer.

That’s why I want Fragment Practice’s tone to be less about
“the technology of the future” and more about

AI considered at human distance.

The combination of orange and purple is a small anchor for me,
so I don’t forget that distance.

Slowly aligning colors and structure

Around Fragment and Fragment Practice,
there will probably be more and more “fragments”:
ZINEs, prototypes, working notes from real projects.

For each of them, I don’t want to force a choice
between:

  • the orange side — stories of life, emotion, and background, and
  • the purple side — structure, protocols, and code.

Instead, I’d like to treat them as gradients.

Choosing colors looks, on the surface,
like a tiny design decision.
But for me, it’s also about deciding where along the line between humans and AI I’m choosing to think from.

The orange of the Setouchi morning.
The quiet purple beyond the screen.

Moving back and forth between those two,
I’d like to keep tending Fragment Practice
as a small warm greenhouse for thought.

Related reading

Read next

Related arguments, records and observations to continue with.

Work from November 2025

Personal essay

2 min read

Urban rhythm, local rhythm

An essay from 2025: Tokyo’s density and Takamatsu’s breathing room, seen through the colors and margins of a canvas.

Personal essayHuman-AI work

Published 16 Nov 2026

Retrospective

4 min read

Where Does the Conversation Continue?

In 2025, core, persona and meta were a way to describe context and shifts in a conversation. Returning to those designs reveals both a connection and…

RetrospectiveHuman-AI workOperating practiceResponsibility boundaries