Lab preview
Onyx
Protect questions and company knowledge before model inference.
Integration maintained by Innovius
Product previews come from the product teams; the illustration shows the ShinrAI workflow. Click an image to explore.
Ask about your company knowledge
Ask a question about a fictional customer document in Onyx. Route the model call through the ShinrAI relay and inspect how the question and retrieved passages are protected before they reach the model.
What is protected
The customer-run relay protects onward model requests, including retrieved context. Originals remain in Onyx. Use local embeddings and route all selected models through the relay.
Tested versionsOnyx 4.7.1 · ShinrAI relay 0.1.0
Install and connect
Get your ShinrAI key
Create an account or sign in. Choose your plan and create an API key in Dashboard, or use a key you already saved. Existing key limits apply.
Install and connect
Start the supplied relay container with your ShinrAI key and existing model credentials. Add it as an Onyx custom inference provider and select the protected models for normal and auxiliary calls.
Verify protection
Run the included connection test and synthetic example. Inspect the destination: the original name and email should be replaced, and no restoration mapping should be included.
Contact Max Mustermann: max.mustermann@example.org.
If protection is unavailable, the affected operation stops. Check your key, allowance and connection, then retry using the product’s instructions.
Free connector, metered service
Connector source and licenses are included. ShinrAI usage is billed through your account. Direct installation is available independently of marketplace approval; no marketplace listing is claimed.
Your privacy controls remain essential: detection depends on the data. Test representative examples before using the integration with customer content.