Knowledge & Agent Platform
A grounded AI knowledge system where every fact has a source.
In production — internal
General-purpose AI models guess when they should cite. This platform grounds AI output in a curated knowledge base: a structured knowledge tree spanning fifteen domains, entity data providers, and a fact layer with source approval, provenance tracking and audit trails — so every claim the system makes can be traced to where it came from.
What’s inside
- Structured knowledge tree across fifteen domains
- Entity providers that assemble fact packages for AI prompts
- Source approval workflow with provenance and batch audit trails
- Flagging and review so bad data can be caught and corrected
- Local model serving with a streaming admin interface
Stack
PHP
MariaDB
Typesense
Ollama (local LLMs)
What it proves: private, grounded AI is practical — the same architecture we deploy for firms that need AI answers backed by their own vetted information.