Akasha is a local-first, concept-oriented operating system.
It begins with a simple observation:
Human beings rarely think in files.
We think in concepts.
A remembered sentence.
A quotation.
A place on a map.
A field observation.
A scientific hypothesis.
A recipe.
A story.
Modern software organizes information into documents, records, tables, and pages.
Akasha organizes meaning.
The easiest way to understand Akasha is not to read about it.
It's to explore it.
🌍 Akashic Archives
Explore a live semantic knowledge space in your browser.
Start in the archives themselves:
- 📚 Reading Hall — wander the knowledge space the way you would a library, one concept opening onto the next
- ✨ Semantic Cosmos Tour — board a guided flight through the semantic space itself, and watch meaning arrange itself into constellations
Or step into a domain portal:
- 🍽 Akashic Kitchen
- 🌎 World Flavors
- 🥘 Regional Recipes
- 🧀 Pairing Explorer
The semantic space is one.
Each portal is a different projection of it—a different way to enjoy the same intellectual exploration.
More concept worlds are coming.
→ https://akashicarchives.net/
They simply call them Share.
When you tap the Share button, you are not thinking about files or data structures.
You are choosing what you want to do with a photo, a document, or a piece of information.
Akasha extends that same idea—
from files and applications
to concepts themselves.
Download the seed from Releases, then:
mkdir akasha
cd akasha
python akasha_seeds_seeds12.pyThe seed expands into a complete local semantic environment.
From the second launch onward:
python akasha.pyNo cloud.
No mandatory infrastructure.
Runs on Linux, macOS, Windows, Raspberry Pi—even iPad.
Unlike traditional software, Akasha treats concepts as the primary unit rather than files, tables, or documents.
Its architecture is built around a few simple principles:
- 🧠 Concepts before implementation
- ⚛️ Three primitives — Atom, Link, and overlapping Set — beneath everything else
- 🔗 Semantic graphs instead of isolated documents
- 🧩 Plug-in Concept Models
- 🏗 World, Workbench, and Workflow on the same substrate as the knowledge they act on
- 🤝 Humans, LLMs, programs, and sensors as peers in one semantic world
- 🔌 Native MCP integration
- 🌍 Browser portals generated from the same semantic space
- 🌱 Local-first, and able to keep working when capabilities retreat
- 📦 Single-file Seeds deployment
- 🪶 MIT licensed
One graph.
One memory substrate.
Many worlds.
The ideas are older than the code. Most of them were formed roughly thirty years ago in knowledge engineering; the 2026 implementation is the point at which hardware, networks, and LLMs let them run.
📖 New to Akasha?
- Explore Akashic Archives
- Run Akasha Seeds
- Read the Quick Start
- Try the Cookbook
Two documents explain why Akasha was designed this way—not just how to use it.
The readable account of the design. It covers:
- Concept-Oriented Computing
- Atom, Link, Set — and why Sets overlap
- Scope and Projection
- Semantic Pipelines
- Concept Models, definition and observation,
dive - Operand · Operator · Agent
- The Semantic Substrate, Harmonia, Silica, and the semantic macro-cache
- A world that changes: telemetry, observation, action
- World, Workbench, Workflow, Chronon
- Society
- Seeds, retreat, and continuity
The canonical architectural reference: the original design intent, kept separate from implementation history, with guardrails for reading the project's past. Where the White Paper and this document differ, this document governs.
Akasha is not trying to become another note-taking application.
Nor another database.
Nor another AI framework.
It is an experiment in treating concepts as computational objects.
The substrate has been built, and it has survived reality.
Now it is being used—
by people,
by LLMs,
by a machine learning plants inside a semantic world,
and eventually,
by the physical world itself.
The expedition is only beginning.
MIT License