SOGAMBOT
A deployed university assistant built from messy institutional documents and retrieval workflows.
A university RAG chatbot that turns scattered Sogang documents and administrative knowledge into a searchable assistant.
problem
University information is scattered across pages, notices, PDFs, and local institutional knowledge that users cannot search reliably.
why it matters
It grounds the agent-readable-knowledge thesis in a real user-facing deployment, not just a research prototype.
maker note
This is the product version of the same pattern: take institutional knowledge that nobody can easily search, then make it usable.
what I built
- • RAG assistant workflow
- • Public-facing chatbot/product surface
- • Operational improvements around retrieval and answer quality
evidence
- • Deployed public artifact
- • Connected to DeepFountain/Sogang chatbot work
current status
Deployed system; used as applied evidence for the knowledge-substrate pattern.
next step
Use anonymized usage and failure patterns to improve retrieval and interaction design.