100% Local AI Easy Button
AirgapAI is a fully isolated 100% local ChatGPT-like (but more capable) solution, enabling users to leverage their own data securely, at a significantly lower price point than MS Copilot / ChatGPT. AirgapAI only costs $3-4 per user per month versus MS Copilot $20-30 per user per month
The world thinks in files. We think in Ideas®
Leading AI Productivity Solution Feature Comparison | ||||
| AirgapAI | ChatGPT | MS Copilot | ||
|---|---|---|---|---|
| Price | ≈$3/user/month | $20/user/month | $30/user/month | |
| One-Time Perpetual Licences | ✓ Yes | ✗ | ✗ | |
| Runs on External 3rd Party Server | ✓ No | ✗ Yes | ✗ Yes | |
| No Annual Maintenance Fee | ✓ Yes | ✗ | ✗ | |
| On-Prem | ✓ Yes | ✗ | ✗ | |
| Use Your Proprietary Data | ✓ Yes | ✗ | ✓ Yes | |
| Modular Component Data Governance | ✓ Yes | ✗ | ✗ | |
| Patented Data Ingestion | ✓ Yes | ✗ | ✗ | |
| 78X LLM RAG Accuracy via Blockify® | ✓ Yes | ✗ | ✗ | |
| 100% Local to Device | ✓ Yes | ✗ | ✗ | |
| Role Based Data Provisioning | ✓ Yes | ✗ | ✗ | |
| Legacy PC Support | ✓ Yes | ✗ | ✗ | |
| AI PC Support | ✓ Yes | ✗ | ✓ Yes | |
| Own Your Data | ✓ Yes | ◐ | ◐ | |
| Control of Application and Model Upgrades | ✓ Yes | ✗ | ✗ | |
| Use any Open Source LLM | ✓ Yes | ✗ | ✗ | |
| Control When You Upgrade Your LLM | ✓ Yes | ✗ | ✗ | |
| Use a Fine-Tuned LLM | ✓ Yes | ✗ | ✗ | |
| Quick Start Workflows by Role and Persona | ✓ Yes | ✗ | ✗ | |
| Chat with Multiple AI Personas (Entourage Mode) | ✓ Yes | ✗ | ✗ | |
| Integrates with MS Office | Future* | ✗ | ✓ Yes | |
AirgapAI System Requirements
For users who have the latest generation of hardware with either an integrated GPU or dedicated GPU, AirgapAI offers a large language model inference capability that allows a user to run common open source LLMs 100% locally. Depending on the performance of the hardware, different speeds of LLM output will be experienced (better hardware = faster generation). Running a LLAMA 3.2 3B Default Model on Intel HD Integrated Graphics with 8GB vRAM results in a generation speed of approx. 100+ words per minute as tested on the Intel® Core™ Ultra 7 165U, vPro®.
For users who do not have latest generation hardware with an integrated GPU, dedicated GPU, AirgapAI offers a CPU based vector search capability called Rapid Answers, which can run on any modern CPU (2016 or later).
Recommended/Required specifications (assumes device is not running other compute intensive tasks):
- Large Language Model Inference
- LLAMA 3.2 3B Default Model – Intel HD Integrated Graphics with 8GB vRAM (minimum)
- LLAMA 3.2 1B Default Model – Intel HD Integrated Graphics with 8GB vRAM (minimum)
- GEMMA 9B – NVIDIA RTX 5000 or similar from AMD
- LLAMA 3 8B – NVIDIA RTX 5000 or similar from AMD
- Rapid Answers IdeaBlock Vector Search
- Any Modern CPU 2016 or later