AI Screenshot Analysis

Ask any question about your team's screenshots and get instant AI-powered answers. The AI can identify applications, detect activity patterns, find specific content, and analyze work habits.

Overview

AI Screenshot Analysis transforms how you review time tracking screenshots. Instead of manually scrolling through hundreds of images, simply ask a question like "Was this person watching YouTube?" and get instant, intelligent answers with confidence scores.

Key Capabilities

• Natural Language Questions: Ask anything about what's shown in screenshots
• Application Detection: Identify which applications and websites were being used
• Activity Classification: Determine if activities appear productive or recreational
• Pattern Detection: Find repetitive behavior or suspicious patterns
• Content Search: Look for specific text, websites, or visual elements
• Work Summaries: Get AI-generated summaries of work activities

Getting Started

Who Can Use It

AI Screenshot Analysis is available to:

• All team members for their own screenshots
• Managers and Supervisors for team member screenshots they have permission to view
• Organization Managers and Super Admins for any team member

Token Requirements

Screenshot analysis uses AI tokens:

• 1 token = 100 screenshots analyzed
• Purchase tokens from the Billing page
• Your token balance is shown before starting any analysis

Starting an Analysis

Step 1: Access the Screenshots Page

Navigate to Screenshots from the main menu to view time tracking screenshots.

Step 2: Click "Analyze Screenshots"

Click the Analyze Screenshots button in the top-right corner to open the analysis modal.

Step 3: Enter Your Question

Type your question in natural language. Be specific about what you're looking for:

Good questions:

• "Was this person watching YouTube or other video sites?"
• "Find screenshots showing social media usage"
• "What applications were being used most?"
• "Flag screenshots that don't look like work-related"

Too vague:

• "Analyze these" (What should the AI look for?)
• "Is this good?" (Good in what way?)

Step 4: Select Date Range

Choose the date range of screenshots to analyze. The wider the range, the more screenshots will be processed and the more tokens required.

Step 5: Select User (Optional)

If you have permission to view other team members' screenshots, select which user's screenshots to analyze.

Step 6: Choose Visibility

• Personal: Only you can see the analysis results
• Team: Managers and admins can also view the results

Step 7: Start Analysis

Click Start Analysis. The system will:

1. Queue your analysis job
2. Process screenshots using AI
3. Notify you when results are ready

Example Questions

Productivity Monitoring

• "Is this productive work or distraction?"
• "How much time was spent on work vs entertainment?"
• "Flag any screenshots showing games or streaming sites"
• "Were there long periods of inactivity?"

Application Usage

• "What applications were being used?"
• "How much time was spent in code editors vs browsers?"
• "Find screenshots showing Slack or Teams usage"
• "Which development tools appear most often?"

Content Detection

• "Find screenshots containing spreadsheets"
• "Look for any social media - Facebook, Twitter, LinkedIn"
• "Were there any video calls or meetings?"
• "Find screenshots showing email clients"

Activity Patterns

• "Summarize the work activity for this period"
• "Are there patterns of repeated idle screens?"
• "Was the same screen showing for extended periods?"
• "Identify any breaks or non-work activities"

Understanding Results

Analysis Jobs Page

After starting an analysis, visit Screenshots → View Analysis Jobs to see:

• Status: Queued, Processing, Complete, or Error
• Results Count: How many screenshots matched your query
• Created Date: When the analysis was requested

Results Page

Click View Results on a completed job to see:

AI Summary

A natural language summary of what the AI found across all screenshots analyzed.

Pattern Analysis

Smart detection of patterns in the screenshots:

• Freshness Ratio: Percentage of unique screen states (higher = more varied activity)
• Unique States: How many distinct screen configurations were detected
• Largest Cluster: Size of the biggest group of similar screenshots
• API Calls Saved: How many redundant analyses were skipped due to clustering

Pattern Verdict

AI-generated assessment of any concerning patterns, such as:

• Unusually repetitive screens
• Extended periods of inactivity
• Potential idle time with no screen changes

Screenshot Results

Each screenshot includes:

• Thumbnail: Preview of the screenshot
• Confidence Score: How confident the AI is in its assessment (0-100%)
• Comment: AI's description of what's shown and how it relates to your query
• Timestamp: When the screenshot was captured

Download Results

Click Download JSON to export complete analysis results for your records or further analysis.

Token System

Understanding Tokens

• AI tokens are required for screenshot analysis
• 1 token = 100 screenshots analyzed
• Tokens are deducted when analysis starts

Checking Your Balance

Your current token balance is displayed in:

• The analysis modal before starting
• The Billing page

Purchasing Tokens

1. Go to Billing from the main menu
2. Click Purchase Tokens
3. Select the amount to purchase
4. Complete payment

Cost Optimization

The AI uses intelligent clustering to reduce costs:

• Similar screenshots are grouped together
• Only unique screen states are sent for full analysis
• You're only charged for actual AI processing

Visibility Settings

Personal Visibility

• Only you can see the analysis and results
• Useful for self-review and personal productivity tracking
• Results won't appear in team reports

Team Visibility

• Managers and admins can view the analysis
• Useful for team oversight and reporting
• Creates a shared record of the analysis

Tips for Best Results

1. Be Specific

The more specific your question, the better the results:

• ✅ "Find screenshots showing YouTube, Netflix, or Twitch"
• ❌ "Find entertainment"

2. Use Date Ranges Wisely

• Start with smaller date ranges to test your queries
• Expand once you've refined your question
• Consider analyzing one day at a time for detailed review

3. Review Pattern Analysis

The pattern metrics can reveal issues your specific question might miss:

• Low freshness ratio might indicate idle time
• Large clusters might show repetitive activities

4. Combine Multiple Analyses

Run different queries to get a complete picture:

1. First: "Summarize work activities"
2. Then: "Find any non-work activities"
3. Finally: "Check for extended idle periods"

Privacy and Security

Data Protection

• Screenshots are only accessed for analysis purposes
• AI analysis doesn't store your screenshot content
• Results are stored securely within your team's data

Access Controls

• You can only analyze screenshots you have permission to view
• Visibility settings control who sees results
• All analysis activities are logged

Troubleshooting

"No results found"

• Verify screenshots exist for the selected date range
• Check that the user has screenshots (not manual time entries)
• Try broadening your question

"Insufficient tokens"

• Purchase additional tokens from the Billing page
• Try a smaller date range to reduce token requirements

Analysis taking too long

• Large date ranges require more processing time
• Check the Analysis Jobs page for status updates
• Results typically complete within a few minutes

Low confidence scores

• Some screenshots may be ambiguous
• Review low-confidence results manually
• The AI provides its best interpretation but isn't always certain

Next Steps

• Learn about time tracking screenshots
• Explore the AI Business Assistant for data queries
• Review time reports and analytics