Omniscient AI

Omniscient AI is a FastAPI-based industrial knowledge assistant for ingesting documents, indexing them, and answering questions with retrieval-augmented generation (RAG). The project also includes demo enterprise features for maintenance ticketing and compliance checking for hackathon-style demonstrations.

What this project includes

• Multi-file document upload and ingestion
• Document parsing for PDFs, DOCX, XLSX, CSV, text, images, and more
• Vector search with Qdrant
• Graph-aware retrieval support for Neo4j
• Chat UI for asking questions over indexed documents
• Demo maintenance ticket creation workflow
• Demo compliance checklist evaluation for SOP/manual-style content

Requirements

• Python 3.11+
• pip
• A working environment for the following optional integrations:
  • Qdrant
  • Neo4j
  • Google Gemini API key

Quick start (Windows PowerShell)

1. Clone the repository and enter the project folder

cd C:\path\to\omniscient

2. Create and activate a virtual environment

python -m venv .venv
.\.venv\Scripts\Activate.ps1

3. Install dependencies

pip install -r requirements.txt

4. Create a .env file in the project root

Example:

QDRANT_URL=
QDRANT_API_KEY=
QDRANT_COLLECTION=
EMBEDDING_MODEL=
EMBEDDING_BATCH_SIZE=
CHUNK_SIZE=
CHUNK_OVERLAP=
UPLOAD_DIR=storage/uploads
TEMP_DIR=storage/temp
STATUS_FILE=storage/status/documents.json
MAX_UPLOAD_SIZE_MB=
APP_HOST=127.0.0.1
APP_PORT=8000
GOOGLE_API_KEY=
GEMINI_MODEL=
NEO4J_URI=
NEO4J_USER=
NEO4J_PASSWORD=
NEO4J_DATABASE=
AURA_INSTANCEID=
AURA_INSTANCENAME=

5. Run the app

python -m uvicorn app.main:app --reload

6. Open the app in your browser

• Admin/upload page: http://127.0.0.1:8000/
• Chat page: http://127.0.0.1:8000/chat
• Demo tickets admin page: http://127.0.0.1:8000/admin/tickets

Demo enterprise features

• After an AI response, the app may show Suggested Enterprise Actions.
• If the answer looks maintenance-related, a Create Ticket action appears.
• If the retrieved context looks like an SOP/manual/policy, a Check Compliance action appears.
• Tickets are stored locally in storage/demo_tickets.json.
• Compliance rules are stored in storage/compliance_rules.json.

Project structure

• app/main.py — FastAPI app entry point
• app/api/ — API routes for chat, admin, history, and tickets
• app/services/ — business logic for chat, ingestion, tickets, and compliance
• app/utils/templates/ — HTML templates for the UI
• app/static/ — CSS and JavaScript assets
• tests/ — unit tests for the core services

Running tests

pytest -q

Notes

• The app can run locally without a full production backend, but document ingestion and chat answers depend on your configured Qdrant, Gemini, and Neo4j services.
• The ticketing and compliance features are intentionally demo-focused and modular so they can later be replaced with real enterprise integrations.