# Utopia

An offline, self-hosted enterprise knowledge substrate that builds a bitemporal knowledge graph with ontology and conflict detection, so agents can reason over how understanding changed over time.

Utopia is an enterprise knowledge substrate that learns passively from incoming material and governs itself, built by DeepLethe. Where a conventional knowledge graph or vector store aims to represent present knowledge, Utopia puts time awareness and ontology into its base layer, so conflict detection, reasoning and decision-making run against an ontology that evolves as new material arrives.

It is built as a bitemporal knowledge graph, so that when a decision is reviewed later, the system can reconstruct the full course of reasoning and the evidence it rested on at the time. It deploys offline, on hardware the operator controls, aiming to give a company a knowledge foundation, a decision core its agents can trust, and a compliance audit trail.

It is written in Rust and uses pgvector and PostgreSQL, and the project positions itself as a different route to enterprise intelligence built from knowledge governance upward, rather than an open-source take on existing data-fusion platforms.

Website: https://github.com/deeplethe/utopia