A user-owned, intent-native computing environment in which applications become implementation details, interfaces are composed for the task at hand, and agents can safely reshape the environment itself.
Project status: RESEARCH / DESIGN PROPOSAL — This repository contains research, design artifacts, and a manifesto. There is no running code.
Public research project. The documents describe an evolving proposal, not a finished system. Claims are labeled by maturity and substantial changes should remain reviewable in version control.
Documents in this repository use these labels to distinguish maturity:
| Label | Meaning |
|---|---|
VISION |
Aspirational direction; not yet validated or specified |
RESEARCH |
Investigation of prior art, landscape, or open questions |
DESIGN PROPOSAL |
Concrete proposal subject to revision through evidence and argument |
IMPLEMENTATION |
Specified and ready for (or under) implementation |
- MANIFESTO.md — The project's vision, principles, and boundaries
- roadmap.md — Phased plan from research to prototype
- design/current-direction.md — Current synthesis: multimodal interaction, dependable workflows, precedents, local inference, and just-in-time hardware
- research/poc-foundation-recommendation.md — Evidence-backed Linux shell, agent, and generative-UI PoC recommendation
intent-centric-computing/
├── README.md # This file
├── MANIFESTO.md # Vision, principles, and invitation
├── LICENSE # Apache-2.0
├── CONTRIBUTING.md # How to get involved
├── GOVERNANCE.md # Governance model (provisional)
├── roadmap.md # Phased roadmap
├── articles/
│ └── linkedin-beyond-apps.md # LinkedIn article draft
├── research/
│ ├── landscape.md # Prior-art map
│ ├── conceptual-ladder.md # Agent-operated → agent-modifiable → intent-native
│ ├── poc-foundation-recommendation.md # Linux/UI foundation research and PoC path
│ └── sources.md # Cited sources with verification status
├── design/
│ ├── philosophy.md # Core philosophy and models
│ ├── architecture.md # High-level architecture
│ ├── principles.md # Foundational principles
│ ├── current-direction.md # Latest design synthesis and prototype implications
│ └── open-questions.md # Unresolved questions
└── scripts/
└── verify-repo.sh # Repository structure verification
Computing today is organized around paradigms designed by vendors: applications, windows, menus, notifications. Even on open platforms like Linux, practically reshaping your computing environment requires deep expertise.
Large language models change a key constraint: translating human intention into machine procedures is becoming feasible at a general level. This project explores what personal computing looks like when intent — not applications — is the primary abstraction.
We define three levels:
- Agent-operated — AI uses existing apps on your behalf
- Agent-modifiable — AI can modify apps and system behavior
- Intent-native — your desired experience is the starting point; the system discovers how to realize it
This project targets level 3. See the manifesto for the full vision and research/conceptual-ladder.md for the detailed framework.
This project builds on ideas from Omarchy, Ink & Switch, Brain, Open Interpreter, OpenDAN, and the historical lineage of Kay, Goldberg, and Licklider. See research/landscape.md for the full analysis.
See CONTRIBUTING.md. This project is in its earliest research phase — discussion, critique, and ideas are the most valuable contributions right now.