Local AI with Ollama (Desktop Agent)
Run the XHack Agent entirely on your own machine by connecting a local Ollama model — no data leaves your network, no per-token cost, and full operation in air-gapped and classified environments.
Keep everything on your machine
Some assessments involve targets so sensitive that data simply cannot leave the network. The XHack Agent — the desktop application — supports this directly by connecting to a local Ollama model. When the Agent runs on a local model, all AI processing happens on your machine: nothing is sent to an external API, nothing is logged on a remote server, and no internet connection is required.
That makes the Agent suitable for air-gapped environments, classified networks, and privacy-sensitive engagements where cloud AI is not permitted — with the same autonomous workflow you'd use online.
Local AI lives in the desktop Agent, not the web platform. For connecting your own cloud provider key on the web app, see Bring Your Own Key.
Unlimited local usage, no per-token cost
Cloud models bill per token, which adds up across a long engagement that generates a large volume of analysis. A local Ollama model has no per-request cost — run as many queries as you need. It's ideal for large-scale assessments, continuous testing, and training exercises. Keep your cloud budget for the operations that need a frontier model, and run everything else locally.
Choose your local model
Pull whichever Ollama models you want on your machine, and the Agent lists them for selection. Pick the one that fits the task — a smaller model for fast, cheap iteration; a larger one when you need more capability — and switch between them without losing your session context.
Setting it up (all in the app)
The Agent is a fully graphical desktop application — there is no command line involved.
- Install Ollama on your machine and pull a model.
- Download and open the XHack Agent, then sign in.
- In the Agent's model settings, choose your local Ollama model. The Agent detects what's available and presents it for selection.
That's it — from that point the Agent runs locally, fully offline if you want it to.
Enterprise & air-gapped deployment
For teams with strict compliance requirements, local AI lets you deploy approved models on internal hardware and keep every byte of assessment data inside your own perimeter — without giving up the Agent's autonomous, AI-driven security workflow.
Ready to get started?
Experience this feature firsthand and see how it can enhance your security operations.
Download the Agent