What Is Dots?

Dots is a knowledge platform for capturing what you and your organization know and keeping it in a form both you and an AI can read. It stores that knowledge as structured records rather than as documents, so one set of rows serves a person reading a table and a model reading context. Structure is also what makes knowledge teachable: a well-formed row is a training example for the Human Intelligence Model, the model Dots trains on human intelligence.

Everything in Dots is built on one primitive: the Dotabase. A Dotabase is a typed collection of rows — each row a title, a body, a set of declared properties, and a place in a knowledge graph. You read it as a table, walk it as a graph, or hand it to an AI as context, all from the same underlying rows.

You reach your knowledge three ways: the web app, the dots command-line interface, and any AI agent connected over the Model Context Protocol. All three are front-ends onto the same data, so a row written through one surface is readable from the other two without an export or a sync step.

Key Features

• Typed rows under a schema. A Dotabase declares which properties its rows may hold, and a write naming an undeclared property is refused rather than stored.
• A workspace-wide knowledge graph. Typed, directed edges connect rows, and a row's edges are listable from either end.
• The DIIICE taxonomy. Six knowledge types — Data, Intelligence, Instructions, Ideas, Context, and Examples — say what kind of knowledge a row is.
• Views as saved lenses. The same rows render as a table, a board, a graph, or any of the ten built-in view types; switching lenses never changes a row.
• Foundational Dotabases from the start. Memory Bank, Conversations, Artifacts, and HUMAN.md ship with a workspace, alongside the internal Dotabases the platform seeds for its own capabilities.
• Chat against your own knowledge. Conversations run over your rows, and you pick the context each one reads.
• Agent access over MCP. A connected coding agent gets tools to search the workspace, list a Dotabase's rows, and create new ones.

Get Started

• Quickstart — nothing running yet: bring an instance up and go through first run.
• Your First Dotabase — build one end to end from the CLI: rows, an edge, and the graph view.
• Quickstart: Agents — start from a coding agent rather than the app.
• Quickstart: Teams — start by bringing colleagues into a shared workspace.

Resources

• How Dotabases Work — the primitive in full: schema, behaviors, edges, views, and the Operation Manual.
• DOTs And The Knowledge Graph — edge types and how a graph is walked.
• DIIICE — where a row's type comes from, and what it is good for.
• Views And Lenses — every view type and its job.
• Foundational And Internal Dotabases — what a new workspace already contains.
• General Overview — navigating the web app itself.