wtf is semantica?

Semantica is an open source infrastructure layer that gives AI agents a structured, explainable memory instead of just a vector index. Most AI agents today store information as embeddings, which are good for finding similar text but cannot explain why a decision was made or show a clear trail from data to conclusion. Semantica addresses that by building what it calls a Context Graph, a knowledge graph that organizes entities, relationships, and decisions in a way that can be queried, audited, and traced back to its source. It sits underneath whatever language model, vector store, or agent framework you are already using, and it does not require an LLM to build the graph, reason over it, or track where facts came from. This matters most in regulated industries like finance, healthcare, legal, and government, where an AI system's decisions may need to survive a regulator asking why something happened months later. The project targets several kinds of teams: AI platform teams building agents that make consequential decisions, data teams working with Databricks or Snowflake who want to turn existing warehouse tables into a governed knowledge graph without exporting the data elsewhere, and compliance or audit teams who need traceable answers rather than a black box. Semantica provides tools for decision intelligence, where every decision is tracked as a searchable, traceable object, as well as governance features like conflict detection and compliance rule checking, deterministic reasoning methods such as forward chaining and SPARQL queries, and a full pipeline for ingesting data from multiple sources, extracting entities and relationships, and merging duplicates. It supports several graph storage backends, including RDF systems like Oxigraph and Neo4j-style labeled property graphs, so teams are not locked into one vendor. It is a Python project, installable with pip, released under the MIT license, and requires Python 3.8 or newer. The full README is longer than what was shown.