Aller au contenu principal

Admin Console Guide

The Noxys admin console is your central command center for managing policies, users, monitoring AI usage, and ensuring compliance across your organization.

Prerequisites​

  • Admin role (Viewers have read-only access to most features)
  • Browser with JavaScript enabled
  • Access to your Noxys instance URL (e.g., https://noxys.cloud)

Dashboard Overview​

The Dashboard is your first view when logging in. It provides real-time visibility into AI usage patterns and policy effectiveness.

Dashboard Components​

What you see: Line chart showing interaction count over time (last 7 days).

What it tells you:

  • Growing trend = increasing AI adoption
  • Spikes = specific departments or projects using more AI
  • Baseline = normal usage pattern to detect anomalies

Actions:

  • Click any point on chart to drill down to interactions for that time
  • Toggle between line and bar chart views
  • Export data as CSV

2. Risk Score Distribution​

What you see: Histogram of risk scores (0-1 scale).

What it tells you:

  • Left side (low scores): Safe, approved usage
  • Right side (high scores): Risk-Signal-heavy interactions needing attention
  • Peak position: Where most interactions fall (typical risk profile)

Interpretation:

  • Tall left bar, small right tail = healthy (most usage is safe)
  • Flat/wide distribution = mixed usage (need better policies)
  • Right-heavy = concerning (lots of sensitive data exposure)

3. Top Platforms​

What you see: Bar chart ranking AI platforms by interaction count.

Example:

ChatGPT      [████████████████████] 45%
Claude [████████████] 28%
Gemini [████████] 18%
DeepSeek [██] 6%
Other [█] 3%

Actions:

  • Click platform name to filter all dashboard data to that platform
  • Identify which services are most used (for policy prioritization)

4. Policy Effectiveness​

What you see: Table showing each enabled policy with:

  • Policy name
  • Triggered count (last 7 days)
  • Action type (Block/Coach/Log)
  • Last triggered

Example:

PolicyTriggeredActionLast Triggered
Block PII on ChatGPT127Block5 mins ago
Coach Finance on US Services34Coach2 hours ago
Log all DeepSeek89Log10 mins ago

Actions:

  • Click policy name to edit or view details
  • Toggle enable/disable (without leaving dashboard)
  • Identify underutilized policies for adjustment

5. User Activity​

What you see: Recent user list with last login and status.

Example:

UserEmailRoleLast LoginStatus
Alicealice@acme.frAdmin2 hours agoActive
Bobbob@acme.frViewer3 days agoActive
Carolcarol@acme.frPendingNeverPending

Actions:

  • Click user to invite/reset/remove
  • Identify inactive users who may need offboarding

Dashboard Filters & Customization​

Time Range:

  • Default: Last 7 days
  • Options: 24 hours, 7 days, 30 days, custom date range
  • Click date selector to change

Platform Filter:

  • Default: All platforms
  • Click any platform in "Top Platforms" to filter all charts
  • Or use dropdown to select specific platform(s)

Department Filter (if SSO enabled):

  • Default: All departments
  • Filter by Entra ID group, LDAP department, or SAML attribute

Refresh Rate:

  • Real-time (updates every 5 seconds)
  • Auto-pause if you're reading a section (resumes on scroll)

The left sidebar provides access to all major sections:

Dashboard          [current page icon]
├─ Policies
├─ Users
├─ Interactions
├─ Compliance
├─ Audit Log
├─ Settings
└─ Help & Support

At the top of each page, you'll see the current path. Click any breadcrumb to jump back:

Dashboard > Policies > "Block PII on ChatGPT" > Edit
[click "Policies" to go back to policies list]

Search & Filtering​

Most list pages have a search bar and filter options:

Example (Policies page):

[Search policies...] [Enabled ▼] [By Platform ▼] [By Action ▼]
  • Search: Instant full-text search (name, description)
  • Status: Enabled/Disabled/All
  • Platform: chatgpt, claude, gemini, etc.
  • Action: Block/Coach/Log

Policies Section​

Manage all your AI usage policies from this central hub.

View All Policies​

  1. Click Policies in sidebar
  2. You'll see a table with:
    • Policy name
    • Description (if provided)
    • Enabled toggle
    • Priority number
    • Action (Block/Coach/Log)
    • Triggered count (last 7 days)
    • Last modified

Example:

NameEnabledPriorityActionTriggeredModified
Block PII on ChatGPT✓10Block1272 hours ago
Coach Finance on US✓20Coach341 day ago
Log all DeepSeek✗30Log03 days ago

Create a New Policy​

  1. Click + New Policy

  2. Fill out the form:

    FieldInstructionsExample
    NameUnique, descriptive name"Block PII on ChatGPT"
    DescriptionOptional, max 500 chars"Prevents sensitive data sharing on US-based service"
    EnabledToggle to activate now (off by default)Off
    PriorityLower = evaluated first (0-1000)10
    ActionBlock / Coach / LogBlock
  3. Click + Add Condition to add rules:

    • Choose a field (platform_id, risk_score, user_id, etc.)
    • Choose an operator (eq, neq, in, contains, gt, gte, etc.)
    • Enter a value
    • Click Add
  4. Repeat for multiple conditions (all must match)

  5. Click Create Policy

Edit a Policy​

  1. Click a policy name in the list
  2. Click Edit
  3. Modify any field (name, description, conditions, action, priority)
  4. Click Save

The policy takes effect immediately for future interactions.

Toggle Enable/Disable​

  1. Find the policy in the list
  2. Click the Enabled toggle
  3. No confirmation needed (toggle is immediate)

Disabled policies are not evaluated, making this useful for:

  • Testing new policies without blocking users
  • Temporarily pausing a rule for investigation
  • A/B testing policy effectiveness

Delete a Policy​

  1. Click a policy name
  2. Click Delete
  3. Confirm: "Are you sure? This action is permanent."

Important: Deleting a policy does NOT delete historical records. All past interactions remain in the audit log.

Users Section​

Manage team members, roles, and SSO settings.

View All Users​

Click Users to see:

UserEmailRoleLast LoginStatusActions
Alicealice@acme.frAdmin2h agoActive[...]
Bobbob@acme.frViewer3d agoActive[...]
Carolcarol@acme.fr—NeverPending[...]

Status meanings:

  • Active: User has logged in at least once
  • Pending: Invitation sent, user hasn't accepted yet
  • Inactive: User hasn't logged in for 60+ days

Invite Team Members​

  1. Click + Invite User
  2. Enter email addresses (comma-separated for bulk invites)
  3. Select role: Admin or Viewer
  4. Click Send Invitations

What happens next:

  • Email sent with signup link
  • Recipient creates password and logs in
  • Access granted immediately

Change User Role​

  1. Click a user's name
  2. Click the Role dropdown
  3. Select new role: Admin or Viewer
  4. Click Save

The user's permissions update immediately.

Role Permissions:

ActionAdminViewer
View Dashboard✓✓
View Interactions✓✓
View Audit Log✓✗
Create/Edit Policies✓✗
Manage Users✓✗
View Billing✓✗

Remove a User​

  1. Click a user's name
  2. Click Delete
  3. Confirm: "Are you sure?"

What happens:

  • User loses access to dashboard
  • All historical interactions remain (not deleted)
  • User's email becomes available for re-invite

Single Sign-On (SSO) Setup​

For enterprise deployments, Noxys supports:

  • Entra ID (Microsoft Graph API) — departments, groups, device compliance
  • LDAP / LDAPS — on-premise Active Directory
  • SAML 2.0 — Okta, OneLogin, Ping Identity
  • OIDC — Generic OIDC providers

Entra ID Setup​

  1. Go to Settings → Single Sign-On → Entra ID
  2. Click Configure
  3. Enter your Azure Tenant ID
  4. Grant Noxys permission to read user/group data (Entra ID prompt)
  5. Test by logging out and clicking Sign in with Microsoft

After enabled:

  • Users log in with their Microsoft 365 credentials
  • Their department, manager, and group memberships are synced hourly
  • You can create policies based on AD groups (e.g., "Block DeepSeek for Finance team")

LDAP Setup​

  1. Go to Settings → Single Sign-On → LDAP
  2. Enter connection details:
    • Server: ldap.company.com (or ldaps://... for TLS)
    • Port: 389 (LDAP) or 636 (LDAPS)
    • Bind DN: Service account (e.g., cn=noxys,ou=service,dc=company,dc=com)
    • Bind Password: Service account password
    • Base DN: Search root (e.g., dc=company,dc=com)
  3. Click Test Connection
  4. Once successful, click Enable

SAML / OIDC Setup​

Contact support@noxys.eu for guided setup. You'll need:

  • Your identity provider's metadata URL (SAML) or discovery endpoint (OIDC)
  • Noxys will provide entity ID / client ID and callback URLs

Interactions Section​

View and analyze all detected AI interactions.

View Interactions​

  1. Click Interactions in sidebar
  2. You'll see a table with recent AIInteraction events:
TimestampUserPlatformRisk ScoreClassificationsPolicy Decision
2026-03-20 14:32alice@acme.frchatgpt0.95EMAIL, PHONEBlocked
2026-03-20 14:25bob@acme.frclaude0.15EMAILLogged
2026-03-20 14:10carol@acme.frgemini0.00—Logged

Filter Interactions​

Use the filter panel:

  • User: Select specific user or department
  • Platform: Filter by chatgpt, claude, gemini, etc.
  • Risk Score: Min-Max slider (0-1)
  • Action: Block / Coach / Log
  • Date Range: Custom start and end date
  • Has Classifications: Yes / No (only Risk-Signal-containing)

Example filter: Show all Blocked interactions on ChatGPT from Finance team in the last 24 hours.

View Interaction Details​

Click any row to open detailed view:

Interaction ID: f47ac10b-58cc-4372-a567-0e02b2c3d479
Timestamp: 2026-03-20T14:32:00Z
User: alice@acme.fr (Department: Finance)
Platform: ChatGPT
Direction: Outbound (prompt)
Content Hash: a665a45920422f9d417e4867efdc4fb8a04a1f3fff1fa07e998e86f7f7a27ae3
Risk Score: 0.95
Classifications:
- EMAIL (confidence: 0.98) [24-42 chars]
- PHONE (confidence: 0.92) [45-57 chars]
Policy Decision: Block
Policy Name: Block PII on ChatGPT
Policy ID: c8d4e2f1-aaaa-bbbb-cccc-000000000001
Blocked: Yes
Timestamp: 2026-03-20T14:32:02Z

Export Interactions​

  1. Apply filters as needed
  2. Click Export CSV
  3. All filtered interactions downloaded as CSV file

CSV columns: Timestamp, User, Platform, Risk Score, Classifications, Policy Decision, etc.

Use cases:

  • Compliance reporting
  • Incident investigation
  • Trend analysis (import into Excel/BI tool)

Compliance Section​

Demonstrate compliance with regulations and frameworks.

EU AI Act Mapping​

  1. Go to Compliance → EU AI Act

  2. You'll see a checklist of requirements:

    • ✅ Prohibited AI practices (banned models)
    • ✅ High-risk AI (monitored and logged)
    • ✅ Transparency (disclosure logs)
    • ✅ Human oversight (policies enforced)
  3. For each requirement, review:

    • Status: Compliant / Non-compliant / Partial
    • Evidence: Policies, audit log entries, classifications
    • Action: What to do if non-compliant

Generate Compliance Report​

  1. Go to Compliance → Reports
  2. Click + Generate Report
  3. Select scope:
    • Scope: Your entire org, specific department, or date range
    • Standards: EU AI Act, GDPR, ISO 27001, etc.
  4. Click Generate

Report includes:

  • Policy inventory and effectiveness
  • Risk Signals detection statistics
  • Risk score distribution
  • Audit log sample
  • Recommendations for improvement

Export as: PDF or JSON

Data Residency​

  1. Go to Compliance → Data Residency
  2. Confirm your deployment location:
    • SaaS (EU): All data in European data centers (AWS eu-west-1, Azure westeurope, or similar)
    • Self-hosted: On your own infrastructure (Kubernetes, Docker Compose)
  3. Verify database location and backup destinations
  4. For GDPR compliance, confirm that third-party services (e.g., analytics) are also EU-based

Audit Log Section​

Review all administrative actions (immutable record).

View Audit Log​

  1. Click Audit Log in sidebar
  2. You'll see a table with columns:
    • Timestamp: When the action occurred
    • User: Who performed the action
    • Action: What they did (created policy, deleted user, etc.)
    • Resource: What was affected (policy ID, user email, etc.)
    • Status: Success / Failure

Use the filters to find specific events:

User: alice@acme.fr
Action: policy.created
Resource: Block DeepSeek
Date Range: 2026-03-15 to 2026-03-20

Example results:

TimestampUserActionResourceStatus
2026-03-20 10:15alice@acme.frpolicy.created"Block DeepSeek"Success
2026-03-20 10:16bob@acme.frpolicy.enabled"Block DeepSeek"Success
2026-03-20 14:32alice@acme.frpolicy.edited"Block DeepSeek"Success

View Details​

Click any row to see full details (before/after state):

Action: policy.edited
Timestamp: 2026-03-20T14:32:00Z
User: alice@acme.fr
Resource: "Block DeepSeek" (policy-123)

Before:
priority: 20
enabled: false

After:
priority: 10
enabled: true

Export Audit Log​

  1. Apply filters as needed
  2. Click Export CSV

CSV includes all columns. Useful for compliance reports or investigation.

Settings Section​

Configure webhooks, API keys, preferences, and billing.

API Keys​

Generate tokens for programmatic access (CLI, integrations, SDKs).

  1. Go to Settings → API Keys
  2. Click + Generate Key
  3. Enter a description (e.g., "Splunk integration")
  4. Select permissions:
    • Read: View interactions, policies, inventory
    • Write: Create/update policies, send interactions
    • Admin: User management, audit log (use with caution)
  5. Click Generate

Important: Copy the token immediately (you won't see it again). Store securely in your password manager or secret management tool.

To revoke: Click the key → Revoke

Webhooks​

Send real-time alerts to external systems (Slack, Splunk, Datadog, custom endpoints).

  1. Go to Settings → Webhooks
  2. Enter your webhook URL (e.g., https://hooks.slack.com/services/...)
  3. Click Test Webhook (sends a test event)
  4. Select which events to send:
    • All alerts: Every policy decision
    • High risk only: risk_score > 0.8
    • Policy violations: Blocked interactions only
    • Custom: Select specific classifications (Risk Signals types, amounts, etc.)
  5. Click Save

Webhook payload format:

{
"event_type": "policy_violation",
"timestamp": "2026-03-20T14:32:00Z",
"interaction_id": "f47ac10b-58cc-4372-a567-0e02b2c3d479",
"user_email": "alice@acme.fr",
"platform_id": "chatgpt",
"risk_score": 0.95,
"classifications": [
{
"type": "EMAIL",
"confidence": 0.98
}
],
"policy_decision": {
"policy_id": "c8d4e2f1-aaaa-bbbb-cccc-000000000001",
"policy_name": "Block PII on ChatGPT",
"action": "block"
}
}

Theme & Language​

  1. Go to Settings → Preferences
  2. Choose:
    • Theme: Light, Dark, or System (auto)
    • Language: English, French (more coming)
  3. Save

Changes apply immediately.

Billing Portal​

Access Stripe's customer portal to manage subscription:

  1. Go to Billing → Manage Subscription
  2. You're redirected to Stripe
  3. View invoices, update payment method, manage seats

Keyboard Shortcuts​

Quick navigation:

  • G then D → Go to Dashboard
  • G then P → Go to Policies
  • G then U → Go to Users
  • ? → Show all shortcuts

Actions:

  • / → Focus search bar
  • N → New item (policy, user invitation, etc.)
  • Ctrl+S → Save (on edit pages)
  • Escape → Close modal or deselect

Next Steps​


Need help?

  • Email: support@noxys.eu
  • Chat: In-app support (click ? icon)
  • Status: status.noxys.cloud