Dashboards & Reporting

Noxys provides real-time dashboards and comprehensive reporting capabilities to monitor AI usage, demonstrate compliance, and identify risks across your organization.

Overview

Built-in Dashboards:

• Executive Summary — high-level metrics
• Detailed Analytics — usage trends and patterns
• Risk Assessment — PII distribution and threat detection
• Compliance Status — regulatory requirements checklist

Report Types:

• Usage Analytics — interaction frequency, platforms, users
• Compliance Reports — EU AI Act, GDPR, ISO 27001 mapping
• Security & Risk — PII types detected, policy violations
• Custom Reports — build your own metrics

Main Dashboard

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

Dashboard Components

1. Interaction Trends (Chart)

What you see: Line chart showing interaction count over the last 7 days.

Metrics:

• Total interactions (Y-axis)
• Time period (X-axis, 6-hour intervals)
• Color-coded by action: Block (red), Coach (yellow), Log (gray)

Use cases:

• Spot usage spikes (e.g., end-of-quarter spike)
• Detect anomalies (sudden drop = service outage or blocked access)
• Forecast capacity needs

Actions:

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

2. Risk Score Distribution (Histogram)

What you see: Horizontal bar chart showing interactions by risk band.

Bands:

• Minimal (0.00-0.10) — gray bar
• Low (0.10-0.30) — green bar
• Medium (0.30-0.70) — yellow bar
• High (0.70-0.90) — orange bar
• Critical (0.90-1.00) — red bar

Interpretation:

• Left-heavy (mostly green/gray) = healthy (most usage is safe)
• Right-heavy (tall red bar) = concerning (lots of high-risk interactions)
• Flat distribution = mixed usage patterns

Actions:

• Click a band to filter all dashboard metrics to that risk range
• Hover over bar for exact count

3. Top Platforms (Bar Chart)

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

Example:

ChatGPT      [██████████████████] 45%  (2,340 interactions)
Claude       [██████████] 28%           (1,456 interactions)
Gemini       [██████] 18%               (934 interactions)
DeepSeek     [█] 6%                     (311 interactions)
Other        [█] 3%                     (156 interactions)

Actions:

• Click any platform to filter all dashboard charts to that service
• Identify which services dominate usage
• Make data residency and compliance decisions based on platform mix

4. Policy Effectiveness (Table)

What you see: List of all enabled policies with trigger statistics.

Policy Name	Triggered	Action	Last Triggered	% of Total
Block PII on ChatGPT	127	Block	5 mins ago	12%
Coach Finance on US	34	Coach	2 hours ago	3%
Log all DeepSeek	89	Log	10 mins ago	8%
Coach on High Risk	215	Coach	1 min ago	20%

Insights:

• High trigger count = Effective enforcement (users getting coached/blocked)
• Zero triggers = Policy may be too strict (never matches) or redundant
• Sudden change = Usage pattern shift (investigate why)

Actions:

• Click policy name to view/edit details
• Identify policies to disable (unused)
• Identify policies to promote to "Block" (if "Coach" working well)

5. User Activity (Table)

What you see: Recent user list with engagement metrics.

User	Email	Last Active	Interactions	Status
Alice	alice@acme.fr	2h ago	156	Active
Bob	bob@acme.fr	3d ago	42	Stale
Carol	carol@acme.fr	Never	0	Pending

Use cases:

• Identify inactive users (haven't logged in 60+ days)
• Spot power users (high interaction counts)
• Detect pending invitations that need follow-up

Actions:

• Click user to view their interactions
• Identify users who need offboarding

Dashboard Filters & Time Range

Time Range Selector (top left):

• Default: Last 7 days
• Options: 24 hours, 7 days, 30 days, 3 months, custom date range
• Click to change — all charts update instantly

Platform Filter:

• Default: All platforms
• Click any platform in "Top Platforms" chart to filter
• Or use dropdown to multi-select platforms

Department Filter (if SSO enabled):

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

Action Filter:

• Show: Block, Coach, Log, or combination
• Default: All actions shown

Risk Range Slider:

• Drag to show only interactions in specific risk band
• Example: Show only "High + Critical" (0.7-1.0)

Dashboard Refresh & Sharing

Refresh Rate:

• Real-time (updates every 5 seconds)
• Auto-pause if you're reading (resumes on scroll)
• Manual refresh: Press Cmd+R (Mac) or Ctrl+R (Windows)

Share Dashboard:

1. Click Share (top right)
2. Select time range to include
3. Choose recipients (email list)
4. Click Send

Recipients get a snapshot report via email (not live dashboard).

Analytics Pages

Interactions Page

Purpose: Deep dive into individual AIInteraction events.

View All Interactions:

1. Click Interactions in sidebar
2. See table of all recent interactions (most recent first)

Columns:

Column	Meaning
Timestamp	When interaction occurred
User	Email of user who triggered
Platform	AI service used
Risk Score	0-1 scale (higher = riskier)
Classifications	PII types detected
Policy Decision	Block / Coach / Log

Filtering:

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

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

Actions:

• Click any row to see detailed interaction record
• Select multiple rows to bulk export
• Export filtered list as CSV

Interactions Detail View

Click any interaction to see full details:

Interaction ID: f47ac10b-58cc-4372-a567-0e02b2c3d479
Timestamp: 2026-03-20T14:32:00Z
User: alice@acme.fr
Department: Finance
Platform: ChatGPT
URL: chat.openai.com/c/abc123
Direction: Outbound (user prompt)
Interaction Type: Prompt

Content Analysis:
  Content Hash: a665a45920422f9d417e4867efdc4fb8a04a1f3fff1fa07e998e86f7f7a27ae3
  Risk Score: 0.95
  Confidence: 0.98

Classifications Detected:
  1. EMAIL: alice@acme.fr (confidence: 0.98)
  2. PHONE: +33612345678 (confidence: 0.92)

Policy Decision:
  Policy ID: c8d4e2f1-aaaa-bbbb-cccc-000000000001
  Policy Name: Block PII on ChatGPT
  Action: Block
  Timestamp: 2026-03-20T14:32:02Z

Timeline:
  14:32:00 - Interaction detected
  14:32:01 - Classifications calculated
  14:32:02 - Policy evaluated
  14:32:02 - Action applied (Blocked)

Related Data:

• User's other interactions (timeline)
• Similar interactions (by platform, user, or PII type)
• Policy audit trail (if policy was modified)

Report Builder

Create custom reports combining metrics, filters, and visualizations.

Quick Reports

Pre-built templates:

1. Weekly Usage Summary

   • Total interactions
   • Top platforms
   • Top users
   • Policy violations (count, by policy)

2. Compliance Status Report

   • EU AI Act checklist
   • GDPR data transfer assessment
   • Data residency confirmation
   • PII detection statistics

3. Risk Assessment Report

   • Risk score distribution (histogram)
   • High-risk interactions (detailed list)
   • Classifications detected (pie chart)
   • Trends (risk over time)

4. Department Deep Dive

   • Select department
   • Usage by platform
   • Risk profile
   • User breakdown

Custom Report Builder

Create a custom report:

1. Go to Reports → + New Report
2. Set Report Name and Description
3. Select Scope:
   • All data
   • Specific date range
   • Specific users
   • Specific platforms
4. Add Visualizations:
   • Metrics (e.g., "Total interactions")
   • Charts (e.g., "Interactions by platform")
   • Tables (e.g., "Top 10 users")
   • Heatmaps (e.g., "Usage by hour of day")
5. Add Filters (optional):
   • Only high-risk interactions
   • Only blocked actions
   • Only Finance department
6. Click Preview to see draft
7. Click Save

Example custom report:

Report: "Finance Team AI Usage Weekly"
Scope: Finance department, last 7 days
Visualizations:
  1. Total interactions (metric): 342
  2. Interactions by platform (pie): ChatGPT 60%, Claude 30%, Other 10%
  3. Risk distribution (histogram): Minimal 70%, Low 20%, Medium 10%
  4. Blocked interactions (table): 7 blocked attempts detail
Filters: Only show interactions with PII

Scheduling Reports

Auto-deliver reports on schedule:

1. Create report (see above)
2. Click Schedule (top right)
3. Set frequency:
   • Daily, Weekly (day+time), Monthly (date+time)
4. Set recipients:
   • Email list (comma-separated)
   • Only Admins
   • Full team
5. Click Save Schedule

Example: "Weekly report every Monday 8 AM sent to security@acme.fr"

Compliance Reports

EU AI Act Compliance

Location: Compliance → EU AI Act

What it covers:

• Article 4: Prohibited AI practices (banned models)
• Article 9: High-risk AI monitoring and logging
• Article 13: Transparency with downstream users
• Article 14: Human oversight and decision logs

Per requirement:

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

Example:

Requirement: Article 4 - Prohibited AI Practices

Status: ✓ Compliant

Evidence:
  - Policy "Block DeepSeek" enabled (prevents Chinese AI use)
  - 89 DeepSeek attempts blocked in last 30 days
  - Audit log shows consistent enforcement

Action: No action needed

GDPR Assessment

Location: Compliance → GDPR

What it covers:

• Data transfers outside EU
• Data processing agreements
• Data subject rights (access, deletion)
• Breach notification

Noxys helps demonstrate:

1. Data minimization: Only hashes sent (not raw data)
2. Purpose limitation: AI monitoring only (not sold or shared)
3. Lawful basis: Legitimate business interest (data security)
4. DPA in place: SCCs with Noxys for data processing

Automated checks:

• Services used in EU vs. outside EU
• Missing DPAs (flagged as gap)
• Data residency confirmation

Generate Compliance Report

Location: Compliance → Reports → + Generate Report

Steps:

1. Select Standards:

   • EU AI Act ✓
   • GDPR ✓
   • ISO 27001 ✓
   • SOC 2 (coming)
   • HIPAA (coming)

2. Select Scope:

   • Organization: Entire tenant
   • Department: Specific department (if SSO enabled)
   • Date Range: Custom date range

3. Report Details:

   • Policy inventory ✓
   • Audit log summary ✓
   • PII detection statistics ✓
   • Risk score distribution ✓
   • Recommendations ✓

4. Output Format:

   • PDF (executive summary + charts)
   • JSON (detailed data)
   • CSV (tables only)

5. Click Generate

Report delivery: Emailed within 5 minutes, also available in console

Data Export

Export Interactions

Quick export:

1. Go to Interactions
2. Apply filters (optional)
3. Click Export CSV (top right)

CSV columns:

• Timestamp
• User Email
• Platform
• Risk Score
• Classification Types
• Classification Count
• Policy Decision
• Policy Name
• Interaction ID

CSV size: Up to 10,000 rows per file (paginated)

Export Audit Log

Location: Audit Log → Export CSV

CSV columns:

• Timestamp
• User Email
• Action (policy.created, user.deleted, etc.)
• Resource Name
• Resource ID
• Details (JSON with before/after state)
• Status (Success / Failure)

Use cases:

• Compliance investigation
• User access review
• Policy change history

API-Based Export

For programmatic access:

# Export interactions for a specific date range
curl -X GET "https://api.noxys.cloud/v1/interactions?start_date=2026-03-01&end_date=2026-03-31" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Accept: application/json" \
  > interactions.json

# Export with filters
curl -X GET "https://api.noxys.cloud/v1/interactions?platform_id=chatgpt&risk_score_min=0.7" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  > high_risk_interactions.json

Query parameters:

• start_date / end_date (YYYY-MM-DD)
• platform_id (chatgpt, claude, etc.)
• user_email (specific user)
• risk_score_min / risk_score_max (0-1)
• action (block, coach, log)

Monitoring & Alerting

Alert Configuration

Location: Settings → Alert Severity

Alert Levels:

Level	Criteria	Example	Webhook Sent
Critical	Block + risk > 0.9	Sensitive data on unapproved service	✓
High	Block + risk > 0.7	PII detection on ChatGPT	✓
Medium	Coach + multiple PII types	Multiple classifications trigger	✓
Low	Log only	Allowed usage	✗

Configuration:

1. Go to Settings → Alert Severity
2. Adjust risk score thresholds
3. Toggle which severity levels trigger webhooks
4. Click Save

Webhook Alerts

Send real-time alerts to:

• Slack
• Microsoft Teams
• Datadog
• Splunk
• Custom HTTPS endpoints

Setup:

1. Go to Settings → Webhooks
2. Click + Add Webhook
3. Enter webhook URL (e.g., Slack incoming webhook)
4. Select event types:
   • All interactions
   • High risk only (> 0.8)
   • Policy violations only
   • Custom (select classifications)
5. Click Test Webhook
6. Click Save

Slack example:

🚨 HIGH RISK INTERACTION BLOCKED

Policy: Block PII on ChatGPT
User: alice@acme.fr
Platform: ChatGPT
Risk Score: 0.95
Classifications: EMAIL, PHONE
Time: 2026-03-20 14:32 UTC

View in Noxys: [Link to interaction]

Performance Metrics

Dashboard KPIs (Key Performance Indicators)

In Dashboard, track these metrics:

KPI	Target	Interpretation
Total Interactions	Growing (AI adoption)	Healthy baseline
Blocked Interactions %	5-15%	Balanced enforcement
Average Risk Score	< 0.30	Safe usage pattern
Policy Match Rate	> 80%	Good policy coverage
User Adoption Rate	> 90%	Strong buy-in

Example healthy profile:

• 1,000 interactions/week
• 8% blocked
• Avg risk 0.22
• 85% match existing policies

Red flags:

• Sharp drop in interactions (might indicate disabling extension)
• Spike in blocking (might indicate too-restrictive policy)
• New services emerging (shadow IT)

Custom Dashboards

Create department-specific dashboards:

1. Go to Dashboards → + New Dashboard
2. Name it: e.g., "Finance Team AI Activity"
3. Add widgets:
   • Interaction trends (Finance only)
   • Top platforms (Finance only)
   • Risk distribution (Finance only)
4. Set permissions:
   • Finance Admins (edit)
   • Finance Viewers (read)
5. Save

Result: Finance team has their own view of their AI usage.

Best Practices

1. Weekly Review

Set a calendar reminder to:

• Check Dashboard for anomalies
• Review Top Platforms for shadow IT
• Look at Policy Effectiveness for adjustments needed

2. Monthly Compliance Report

First Friday of each month:

• Generate Compliance Report (EU AI Act + GDPR)
• Send to Security & Compliance team
• File for annual audit

3. Quarterly Executive Brief

End of each quarter:

• Generate Usage Summary Report
• Highlight trends (AI adoption, policy enforcement)
• Present to leadership

4. Continuous Monitoring

• Dashboard alerts: High-risk interactions to Slack
• Weekly email: Digest of policy violations
• On-demand export: Support team uses for incidents

5. Data Retention

Plan for long-term reporting:

• Default retention: 90 days
• Compliance tier: 1-7 years
• Archived data: Available for export (slow query)

Troubleshooting

Dashboard Not Updating

Problem: Metrics look stale (old timestamps).

Solution:

1. Hard refresh browser: Ctrl+Shift+R (Windows) or Cmd+Shift+R (Mac)
2. Check internet connection
3. Try in incognito/private mode (rules out browser extensions)
4. Contact support if persists

Report Generation Slow

Problem: Report taking >5 minutes to generate.

Solution:

• Reduce date range (e.g., last 30 days instead of 1 year)
• Narrow filters (e.g., single department instead of all)
• Try again off-peak hours (fewer concurrent reports)
• Contact support for custom query optimization

Export CSV Too Large

Problem: CSV file can't open in Excel (>1 million rows).

Solution:

• Export in chunks: 1 month at a time
• Use JSON export instead (supports streaming)
• Open in Google Sheets (unlimited rows)
• Use API to stream data to database

Next Steps

• Admin Console Guide — Full feature reference
• Policies — Manage policy effectiveness
• Compliance Guide — Regulatory requirements

---

Need help?

• Email: support@noxys.eu
• Report templates: Available in console under Report Templates
• BI integration: Contact sales for Tableau, Power BI connectors