VeilAnalytics
Home CSV AI Ollama Mode Text-to-SQL Privacy Audit Pricing Try Live Demo →
🟢 100% Offline AI Stack

Private AI Analytics with
Ollama + DuckDB.

For security teams, defense engineers, and privacy advocates who cannot make a single outbound HTTP call. Run local LLMs on your hardware and query multi-million row datasets in plain English.

⚡ Launch Local Demo Workspace → 📖 View Ollama Connection Guide

The Air-Gapped Local Architecture

When using Ollama mode in VeilAnalytics, network traffic is strictly isolated to localhost. Unplug your Ethernet cable or turn off Wi-Fi — the entire analytics workflow continues unaffected.

# 1. Start your local Ollama instance with CORS enabled for web app access:
OLLAMA_ORIGINS="*" ollama serve

# 2. Pull a high-performance open-source SQL model:
ollama pull llama3.1:8b
# or specialized SQLCoder:
ollama pull sqlcoder:7b

# 3. In VeilAnalytics, switch AI Provider to:
# Provider: Local Ollama (http://localhost:11434)
# Model: llama3.1:8b

Why Ollama + DuckDB Outperforms Cloud Analytics

Zero Cloud Bills
$0 / Query
No token costs, no API rate limits, and no expensive monthly per-seat Snowflake or Tableau licenses.
Air-Gapped Compliance
Strict Compliance
No Business Associate Agreements (BAAs) required because data never leaves your hardware boundary.
Native Speed
Sub-50ms Compute
Vectorized DuckDB C++ execution finishes aggregations before cloud network packets can even handshake.