Introduction

Dominir is a system for storing and organizing your company's knowledge — the decisions made, the relationships between teams, projects, and people, and the context behind them — on hardware you control.

Most tools record transactions: sales closed, tickets resolved, invoices paid. Dominir records meaning: who decided what, what it connects to, and why it was done. That body of structured knowledge is called a CIM — Company Intelligence Management.

What a CIM is

Think of it as a permanent, structured memory for your organization. Every entry is a triplet: a subject, a relationship, and an object. For example:

• ▸Project Atlas → is owned by → Infrastructure team
• ▸Decision: migrate to Rust → was made by → Alice Chen
• ▸Customer Acme Corp → anchors contract → Q3 2025 renewal

These triplets build a graph — a web of connected facts your agents, dashboards, and workflows can query against directly.

Where the data lives

The Kernel is the core of Dominir. It is a program that runs on your own machine or server, written in Rust for reliability and speed. It owns all stored data, manages how records are written and read, and controls what gets archived. It never phones home.

The browser interface is a display layer only. It shows you what the Kernel holds and lets you interact with it. If you close the browser, the Kernel keeps running. Data is never stored in the browser — the browser is just a window.

What "Sovereign" means here

Sovereign means the system runs on infrastructure you own and control. No vendor holds your data. No subscription can lock you out. No terms-of-service change can delete your records. Your company knowledge stays inside your network, on your disks, under your policies.

Trade-offs you should know

• ▸Dominir stores data in an append-only format. Old records are never overwritten, which makes the history reliable — but it means the system periodically needs to compact older records to maintain read speed.
• ▸The graph structure is strict: records must conform to a defined schema. Data that doesn't fit is flagged as a Shadow Petiole and held for manual review, rather than silently dropped or corrupted.