Autonomous Agents
Semantic Document Knowledge Graph
UNTH.AI (Internal Project) · Information Systems

The Problem
Our own notes, guides, and past chats were spread across many different tools. Our team and our AI helpers wasted time hunting for the right piece of information.
What We Built
We built a shared memory that reads all of those notes and guides, remembers what they say, and understands how they connect, so anyone can ask a question and get the exact answer.
How It Works
- It reads through all our documents and past conversations.
- It stores them in a way it can search by meaning, not just by exact words.
- It maps how the different pieces relate to one another.
- When asked a question, it hands back the exact matching information.
What Changed
Our team and our AI helpers now find the right answer quickly, and the helpers give more accurate replies because they are working from real, correct information.
CLIENT:UNTH.AI (Internal Project)
INDUSTRY:Information Systems
TECH STACK:
PostgreSQL + pgvectorExpress 5 (TypeScript)MCP ServerXenova TransformersReact 19 + ViteDocker