On-Prem PoC Setup
Version 1.0.71
Fully local RAG stack - no external cloud calls.
Architecture
Everything runs on a single on-prem machine with isolated services. No outbound network calls are required to run indexing, retrieval, and RAG.
Features
Try out each feature: index, retrieve, and chat.
Index
Upload, chunk, and embed content.
Manage collections, search chunks, and review payloads.
Retrieve
Semantic search and reranking-ready retrieval.
Query your indexed data and inspect context chunks.
RAG chat
Grounded answers with citations.
Test retrieval-augmented generation end-to-end.
Model chat
Pure model chat without retrieval.
Compare behaviors vs grounded responses.
RAG Agent Flow Visualization
Interactive demonstration of the on-prem RAG architecture data flow
Host Machine
Gigabyte G5 KC Laptop
GPU
NVIDIA GeForce RTX 3060
6GB VRAM
CPU
Intel Core i5-10500H
6 cores / 12 threads
RAM
32GB
DDR4
Storage
2TB
SSD
Docker
WSL
Docker Desktop
Ngrok
Internet Tunnel
Secured tunneling
Components
All services hosted locally on the same network.
Cyon PoC service
Proof of concept service for Cyon
Qdrant
Vector store
Postgres
Metadata store
Ollama
LLM runtime
KB service
Knowledge base microservice
Connections
Integrations with other data sources.
Atlassian
Sync documents from your Jira Wiki page
Slack
Sync documents from your Slack conversation
Cyon Support
Sync documents from Cyon Support documents
HelpScout
Sync documents from your HelpScout conversation