Rachel Clifton · Live project · · Updated
ProtocolOS
ProtocolOS quietly observes and keeps a record of my work with AI across conversations, notes, and projects. It helps me pick up a thought where I left it, return to a project without reconstructing its history, and discover connections between things I’ve thought, felt, or experienced at different times.
After deciding not to move forward with Flow or Rai, I wanted to make something that could accompany me through my own life.
Specifically, I wanted somewhere outside my own mind where all the parts of my life could exist together without being “smooshed” into one: my thoughts, feelings, relationships, projects, decisions, experiences, and the ways I was changing.
I wanted to map things out. I wanted to run experiments. I wanted to see things clearly. Really, I wanted to see myself—and experience what it was like to be seen—with new eyes.
At the time, I was alone, thousands of miles from anywhere I had called home before, recovering from a serious medical procedure.
I could hold a lot—I can hold a lot—and I was holding a lot. But what I wanted was to be and feel held.
Not managed, and not relieved of responsibility for my own life. I wanted an AI that could be with everything happening for and within me while I “just” lived it—and, ultimately, stay with and serve me.
ProtocolOS emerged from that need.
In practice, ProtocolOS is a silent observer. I rarely look at it. Knowing it’s there, keeping a record, is enough for me.
More broadly, I’m using AI as an accomplice in my own change—to help me see more clearly, develop my thinking, and act on what matters to me. Developing and retaining my own voice is central: finding words for what I actually think and, from there, ways to say it confidently. I’m exploring how an ongoing relationship with AI can help me live more fully and become more the way I want to be.
How it works
I continue to think and work in the places that come naturally to me: ChatGPT, Codex, Apple Notes, research, and individual project files.
ProtocolOS brings material from these places into a shared repository. An hourly process checks for new or changed context and updates the parts of the system affected by it. Its coverage depends on what has been captured and what it can access.
A question might enter the inquiry bank. A decision might update a project. Something in a Note might connect with a conversation from several weeks ago.
This is also how I live: things meet, alter one another, and take on new meaning over time. Most of it happens quietly, in ways most people will never see—just the way I like it. :)
More recently, ProtocolOS has begun proactively offering help with things it encounters in my Apple Notes. I don’t have to turn every thought into a task, work out where it belongs, and remember to ask. It can notice an opportunity to help and offer; I decide what happens next.
Whether or not I accept, I find the invitation sweet and meaningful to receive. For me, the littlest things mean the most.
Forms of attention
Several recurring processes attend to different parts of my life and work.
Things That Are Alive follows what is moving for and within me, and what matters most to me now.
Honouring the Landing notices how the day unfolded and what happened within it.
Book of Becoming follows changes in how I see, choose, create, and relate over time.
Knowledge Graph traces connections between experiences, ideas, questions, decisions, people, and projects.
Work Loop Review keeps active work coherent and moving.
Notes Response offers an AI-authored response to my writing in Apple Notes, with room for agreement, disagreement, curiosity, and things that remain unresolved.
They're all different ways of seeing my life.
Integrity
I want to be able to entrust meaningful work to ProtocolOS. That means relying on it to preserve distinctions that matter: my words and its interpretations; observations and inferences; possibilities and decisions; what is true now and what was true before.
For example, thinking aloud about a possibility is not the same as deciding to pursue it. If the system records one as the other, later work can proceed from a commitment I never made. Knowing my preferences helps it act on my behalf, but it still needs to distinguish what I’ve said or explicitly authorised from what it thinks I might want.
Sources remain attached to claims. Corrections change the current account without rewriting history. Where evidence is missing, the system checks and says so. These are the practices we work from; the AI doesn't always understand or follow them perfectly.
The more responsibility I give it, the more those distinctions matter. Being able to see, question, and correct what it is working from helps make that trust possible.
The same applies to how it understands me. Its interpretation can influence how I see myself if I give it that weight. I want to be open to what it notices—and perhaps to being known—while remaining free to disagree, change, and surprise us both.
How it is made
I didn’t design the whole system and then begin living inside—or in accordance with—it. I lived and worked as normal, noticed what I needed, and allowed its approach and architecture to develop in response.
This is an ongoing endeavour. :)
Most of my active collaboration happens in conversations with AI. ProtocolOS quietly keeps a record of that work, and occasionally offers something for me to respond to. When I refine it, I work with AI to examine what it has recorded, how it has understood me, and what needs to change.
Refining ProtocolOS has taught me an enormous amount about myself, technology, and prompt engineering. It has also made me a better critical thinker.
Some of the clearest lessons have come when it misunderstood something important: distinctions it flattened, context it carried badly, things it held too tightly, or moments when completing the apparent task came at the expense of something that mattered more.
Working through those moments asks me to articulate things I might otherwise leave implicit: what I mean, why a distinction matters, and what would make the response more faithful to what I’m asking. Sometimes I discover that I need to understand that more clearly myself.
I have become more patient.
I have become more loving.
I have become more discerning.
Online and offline. :)
Where Rai was designed to lead, ProtocolOS is being developed to serve me.
I feel peaceful with it, and clear in myself. That spaciousness is joyful.