What it is
Every enterprise currently faces the same challenge: the fancy, expensive AI agents they have purchased do not understand their business context. Neo4j, Databricks and others have attempted to address this — but they only provide the technology and mechanics to build such context. The content is still something the enterprise has to come up with. It is their business context, at the end of the day.
This asset is a means to create a knowledge graph of an enterprise's business context that can be fed into the long-term memory service or used by an AI agent directly. The stack is Postgres + AGE.
The most interesting part of the challenge is building the ontology, and the most interesting part of ontology building is eliciting tacit knowledge from subject-matter experts. The asset includes a set of knowledge-elicitation activities and a framework that automates adjusting those activities to the enterprise, drawing on an activity repository.
Knowledge-graph creation is currently semi-automated — roughly 70/30. The remaining 30% can be automated too, with another AI agent that runs the elicitation activities against experts from a web or mobile UI. That part is still work in progress.
Architecture
Lifecycle
Knowledge creation
Ontology development
Subject-matter-expert knowledge elicitation
