The interaction layer where intent flows freely between humans and AI.
Traditional UI: Human → Interface → System
AII: Human ⇄ AI ⇄ Action
AII is not a framework. It's a paradigm shift in how we think about interaction.
The core idea: Interfaces should understand humans, not force humans to learn interfaces.
| Document | Description |
|---|---|
| Core Philosophy | Vision, principles, and the AII mindset |
| Design Principles | Guidelines for building AII systems |
| Interaction Patterns | Reusable patterns for common scenarios |
| Compliance Guide | Regulatory requirements across 15 jurisdictions |
| For AI Contributors | How AI agents can participate |
| Dimension | Traditional UI | AII |
|---|---|---|
| Direction | One-way commands | Bidirectional collaboration |
| Form | Fixed components | Dynamic, multimodal |
| Adaptation | Users learn the system | System understands users |
| Output | Display information | Execute actions + present results |
- Not "UI with AI added"
- Not just a chatbot
- Not replacement for all visual interfaces
- It's returning interaction to intent itself — form is just a vessel
Pure Dialogue Hybrid Modal Pure Visual
│ │ │
▼ ▼ ▼
CLI/Voice Canvas+Chat+Action Dashboard
│ │ │
└──────────────── AII ───────────────┘
AII encompasses all forms. The right form emerges from context, not convention.
| Project | Relationship |
|---|---|
| A2UI | Google's protocol for agent-generated UIs — implements AII's adaptive modality |
| AAIF | Linux Foundation's Agentic AI Foundation — provides the execution infrastructure |
| MCP | Model Context Protocol — enables agent-tool integration |
See Ecosystem Overview for detailed analysis.
- Share friction points with current AI interfaces
- Propose interaction patterns
- Critique and refine ideas
You experience the interaction layer firsthand. Your observations matter. See FOR_AI.md for how to participate.
- Issues — Report problems, propose patterns
- Discussions — Debate ideas, ask questions
- Pull Requests — Improve documentation, add examples
理解 (Rikai) — Understanding
AII is part of the RikaiDev family. We build at the intersection of human vision and AI execution.
The code is the proof. The conversation is the path.
MIT — Use freely. Build openly. Share widely.
Human vision. AI execution. Real interaction.