How to Run an AI Chatbot Locally for Compliant Use

Artificial Intelligence chatbots are everywhere, but did you know you can actually run one directly on your own computer without sending any data into the cloud? In this guide, we explain how to set up a chatbot locally, using GPT-OSS as an example.

We keep things simple and beginner-friendly, making it easy for non-technical users. We also highlight why running AI locally is especially valuable for organisations focused on privacy and compliance, including teams who use Oregani to manage AI policies.


What does “run a chatbot locally” mean?

Running a chatbot locally means installing the software and model directly on your computer. Everything happens on your machine:

  • Your prompts and responses never leave your device
  • You decide when and how the AI runs
  • It can even work without an internet connection once installed

This gives you full control and security while still benefiting from the power of AI.


Why run an AI chatbot locally?

Here are the biggest benefits for individuals and teams:

  • Stronger privacy and compliance
    Data never leaves your computer, making it easier to follow rules like GDPR or internal data guidelines.
  • No ongoing costs
    You are not billed per use, as with online services.
  • Works offline
    Useful for secure workplaces or remote areas.

For organisations, this makes local chatbots an excellent way to stay compliant while still exploring the benefits of AI. Oregani supports this by helping teams define which AI tools and setups are safe and approved.


Example: GPT-OSS via LM Studio

One of the easiest ways to get started is by using LM Studio. This is a simple desktop app that lets you download and run models like GPT-OSS. No coding is required.

Steps:

  1. Download LM Studio from lmstudio.ai
  2. Search for GPT-OSS in the app and download it
  3. Start chatting with your AI chatbot safely and privately

Minimum requirements for running locally

You do not need a supercomputer, but some compute is needed. Here is a friendly breakdown of the recommended specifications:

RequirementMinimum SpecificationRecommended for Best Performance
Operating SystemWindows, macOS, or LinuxLatest version of each OS
Memory (RAM)16 GB32 GB+
Processor (CPU)Modern multicore (Intel i7 or AMD Ryzen 7)Newer generation with higher cores
Graphics (GPU)16GB of VRAM60GB of VRAM
Storage15 GB free space20 GB+ to store larger models