Aller au contenu principal

Quickstart

Welcome to Noxys, the first European sovereign AI Firewall. This guide walks you through setup in 15 minutes.

Prerequisites​

Before you start, ensure you have:

  • Browser: Chrome, Edge, Brave, or Firefox (latest version)
  • Administrator access: to your organization's Noxys tenant
  • Internet connection: for cloud-hosted deployment, or local network access for self-hosted

Step 1: Sign Up or Log In​

  1. Navigate to your Noxys instance (e.g., https://noxys.cloud for hosted, or your organization's domain for self-hosted)
  2. If you don't have an account:
    • Click Sign Up
    • Enter your work email address
    • Create a secure password (minimum 8 characters)
    • Confirm your email address via the link sent to your inbox
  3. If you already have an account:
    • Click Log In
    • Enter your email and password

After login, you'll see the Onboarding Wizard (first time only) or go directly to the Dashboard.

Tip: If you're the first user in your organization, you'll be assigned the admin role automatically. You can invite team members later.

Step 2: Install the Browser Extension​

The Noxys browser extension monitors AI usage across 15+ platforms. Follow the instructions for your browser:

Step 3: Configure the Extension​

Once installed, the extension needs to authenticate with your backend.

  1. Click the Noxys icon in your browser toolbar
  2. A popup appears asking for your API token
  3. Go back to the Noxys Dashboard → Settings → API Keys
  4. Copy your default API token (or create a new one)
  5. Paste the token into the extension popup
  6. Click Save

The extension icon will change color:

  • Green: Connected and monitoring
  • Red: Connection error (check token, backend URL)
  • Gray: Offline mode (local policies still enforced)

Note: The extension runs entirely in the browser and sends only SHA-256 hashes of detected interactions to your backend. Raw prompt content is never transmitted.

Step 4: Create Your First Policy​

Policies enforce data protection rules. Let's create one to block PII on ChatGPT.

See Creating Your First Policy for detailed step-by-step instructions.

Step 5: View Your Dashboard​

The Dashboard shows real-time insights into AI usage in your organization.

  1. From the left sidebar, click Dashboard
  2. You'll see:
    • Active AI Services: Which platforms employees are using (ChatGPT, Claude, Gemini, etc.)
    • Recent Interactions: Latest detected AI usage
    • PII Detections: Alerts for sensitive data shared
    • Policy Violations: Interactions blocked or flagged by your policies

Key Sections​

SectionWhat it Shows
AI InventoryAll detected AI platforms, user count, activity trends
Risk SummaryHigh-risk interactions by platform, user, or classification
AlertsReal-time security events (policy violations, unusual activity)
ComplianceEU AI Act mapping, data residency, audit readiness

Step 6: Invite Team Members​

  1. Click Users in the left sidebar
  2. Click + Invite User
  3. Enter team member email(s) and select their role:
    • Admin: Full access (policies, settings, users, audit log)
    • Viewer: Read-only (dashboards, alerts, interactions)
  4. Click Send Invitations
  5. Team members receive an email with a signup link
  6. Once they log in and install the extension, they're included in monitoring

Roles: All users see their own usage. Only admins can create policies, manage team members, and access audit logs.

What Happens Next?​

After setup, Noxys automatically:

  1. Detects AI usage via the browser extension (within 10 seconds of visiting an AI platform)
  2. Classifies content for PII (emails, phone numbers, IBAN, NIR, credit cards, etc.) at Tier 1 (regex, <10ms)
  3. Evaluates policies and blocks/coaches/logs interactions
  4. Logs events to the Audit Log (immutable, retained for 90 days by default)
  5. Generates alerts for policy violations, unusual patterns, or compliance issues

Common First Steps​

Enable Tier 2 Classification (Optional)​

For deeper PII detection (medical terms, legal references, IP addresses, API keys):

  1. Go to Settings → Classification
  2. Toggle Enable Server-Side Classification
  3. This runs Presidio + custom recognizers on the backend (adds ~30ms latency per interaction)

Set Up Webhook Alerts​

To integrate with your SIEM/SOAR:

  1. Go to Settings → Webhooks
  2. Enter your webhook URL (e.g., Slack, Splunk, Datadog)
  3. Click Test Webhook to verify
  4. Select which events to send (all alerts, policy violations, or custom)

Enable SSO (Enterprise)​

For Entra ID / LDAP / SAML:

  1. Go to Settings → Single Sign-On
  2. Select your identity provider
  3. Follow the configuration wizard
  4. Users can now log in with their corporate credentials

Troubleshooting​

Extension Shows Red Icon (Connection Error)​

Problem: Extension can't reach the backend.

Solution:

  • Check your API token is correct (Settings → API Keys)
  • Verify backend URL is correct (should auto-detect, but check Settings → Advanced)
  • If self-hosted, ensure your backend is running and accessible
  • Try logging out and back in to refresh the token

No AI Interactions Detected​

Problem: Dashboard shows no activity even though you're using ChatGPT.

Solution:

  • Confirm the extension is installed and enabled (check toolbar, enable if needed)
  • Refresh the AI platform page (ChatGPT, Claude, etc.)
  • Check browser console (F12) for errors and share with support@noxys.eu
  • Confirm the extension has permission to access the website (Settings → Extensions → Permissions)

Policy Not Blocking​

Problem: A policy is enabled, but interactions aren't being blocked.

Solution:

  • Confirm the policy is enabled (toggle must be on)
  • Check the priority (lower number = higher priority; multiple policies are evaluated in order)
  • Verify the conditions match your interaction (e.g., platform_id matches exactly)
  • Review Action: "Block" prevents submission; "Coach" shows a warning

Can't Invite Users​

Problem: Invitation email not received.

Solution:

  • Check spam/junk folder
  • Verify email address is correct (typos prevent delivery)
  • Confirm the user isn't already in your tenant (they'll see an error)
  • If using SSO, users must be in your Entra ID / LDAP group first

Need Help?​

Estimated time to completion: 15 minutes

Next steps: Create Your First Policy or read the Admin Console to learn more.