A way of working — not an app, not a framework, not a chatbot wrapper. A library-based operating system for orchestrating work between you and AI models across sessions, surfaces, and life domains.
This is the seed. What grows from it is your tree.
╭─────────────────────────╮
│ Insights + Actions │ ← Canopy
│ (surfaced by agents, │ What your system produces.
│ approved by you) │ Briefings, analyses, drafts,
╰────────────┬────────────╯ decisions — through your
│ approval gate.
┌───────────────┼───────────────┐
│ │ │
┌─────┴─────┐ ┌─────┴─────┐ ┌─────┴─────┐
│ Health │ │ Career │ │ Finances │ ← Branches
│ │ │ │ │ │ Life domains.
│ agents │ │ agents │ │ agents │ Each branch has agents,
│ certs │ │ certs │ │ certs │ certifications, data,
│ data │ │ data │ │ data │ and personas.
└─────┬─────┘ └─────┬─────┘ └─────┬─────┘
│ │ │
└───────────────┼───────────────┘
│
╭───────────────┴───────────────╮
│ Processes & Conventions │ ← Trunk
│ Growth and Development Loop │ How work happens.
│ Session Lifecycle │ The way of working that
│ Trust Progression │ makes everything above
│ Schemas & Rules │ possible.
╰───────────────┬───────────────╯
│
╭────────────┴────────────╮
│ Your Data │ ← Personal Data
│ Version-controlled │ Everything your system
│ markdown + CSV │ knows, versioned in git.
╰────────────┬────────────╯
│
╭────────────┴────────────╮
│ LLM Services │ ← Roots
│ Claude, GPT, Llama, │ Any model, any surface.
│ local or cloud │ The system is model-agnostic.
╰─────────────────────────╯
Library-based. Everything is version-controlled markdown. Your processes, agent definitions, domain knowledge, data stores — all plain text in a git repository. No proprietary formats, no vendor lock-in, no database to manage.
Model-agnostic. The seed works with any LLM that can read files and follow instructions. Claude, GPT, Llama, Gemini — the system adapts. You declare your environment; agents consult it.
Trust-governed. Agents start untrusted. They earn autonomy through demonstrated performance tracked in a trust ledger. You control the blast radius of every capability.
Self-improving. The Growth and Development Loop — a continuous dev/ops cycle — means your system gets better through use. Every session is an iteration. Errors and friction are inputs, not waste. You build and execute concurrently.
Session-native. Each conversation has a lifecycle: initialize, scope, work, review, close. Drift detection normalizes what changed between sessions. Nothing falls through the cracks.
The Growth and Development Loop is the growth mechanism:
╭──── Dev ────╮ ╭──── Ops ────╮
│ │ │ │
│ Converse │ │ Detect │
│ Produce │ ──────> │ Sweep │
│ Commit │ │ Triage │
│ │ <────── │ │
╰─────────────╯ ╰─────────────╯
Dev side: You and your AI have a conversation, produce artifacts, and commit changes.
Ops side: The system detects drift, sweeps for health, and triages findings into future work.
The crossover: Dev commits become ops state. Ops findings become dev work. This is how a tree grows — building and sustaining happen simultaneously.
personal-ai-seed/
glossary.md Shared vocabulary — co-created in your first session.
processes.md How work happens.
CLAUDE.md Entry point (Claude Code users).
schemas/ Structural definitions.
.claude/rules/ Enforceable process extracts.
system/ Infrastructure and state.
tools/ Tool integration contracts.
prompts/ Bootstrap for non-Claude surfaces.
dream/ Reflective space.
agents/ Empty — you grow these.
certifications/ Empty — you grow these.
personas/ Empty — you grow these.
data/ Empty — you grow these.
plans/ Empty — you grow these.
examples/ Annotated walkthroughs.
The seed gives you the trunk — processes, schemas, governance, and conventions. The branches are yours to grow. Your first experience is a conversation, not a reading assignment — the planting conversation finds a common language between you and your AI.
- Seed — What you're looking at. The distributable pattern.
- Sapling — A seed with branches already growing. Pre-built agents, data structures, and tool guidance for a specific domain. How one tree shares its growth with others.
- Tree — What grows from a seed or sapling. Your personal AI system.
- Grove — A community of trees. People growing together, sharing patterns.
- Forest — All of us. Everyone building their own Personal AI.
Not everyone starts from scratch. A sapling gives you a head start — pre-built structure in one or more life domains, ready to customize. See saplings/README.md for what's available and how to create your own.
Personal AI means you are building a custom system — your own processes, your own agents, your own domain knowledge, your own agentic workflows. This isn't about prompting better. It's about building a system that compounds your capability over time.
The seed teaches this through doing. Your first session is a conversation that finds a common language. From there, you don't read about the Growth and Development Loop — you live it.
See PLANTING.md for the step-by-step guide from zero to your first working session.
This seed was developed through the Kwaai.ai Personal AI community — a nonprofit R&D and policy lab dedicated to making Personal AI accessible to everyone.
MIT