Local inference

Use desktop security analysis and understand what remains local.

Local inference runs security analysis on the device through the Patronus desktop runtime and its security models.

Prepare the device

Install Patronus using the installation guide.
Complete sign-in or organization enrollment as required for your protection capabilities.
Finish the setup or repair actions shown in the app so its runtime and models are ready.
Run the App quickstart and inspect the resulting audits and runtime outcome.

Hardware requirements depend on the selected analysis and model. Confirm them with Patronus before deployment.

Understand the data path

Local security analysis evaluates content on the device. The AI app can still send that content to its configured provider as part of the original request. Organization governance evidence and operational telemetry are separate data flows, described in Privacy and retention.

Calling the hosted Patronus API sends the submitted content to the service. It does not select the desktop’s local analyzer.

Verify and troubleshoot

Inspect audit details to confirm which analysis ran. A successful model response or installed app alone does not confirm local processing.

If analysis cannot run, follow the app’s runtime or repair status. Investigate a trace without the expected audit; it does not confirm a successful check.

For a deployment that must satisfy specific offline or data-residency requirements, confirm the exact runtime, model distribution, and required service connections with Patronus before rollout.