About AI Analysis
TLOGic's AI Playground uses Google Gemini to analyze operator performance data and provide actionable insights. The AI can identify trends, detect anomalies, and suggest improvements based on the metrics extracted from your TLOG file.
How AI Analysis Works
When you request an AI analysis, TLOGic sends operator performance data to Google's Gemini AI model along with your prompt. The AI analyzes patterns in the data and returns natural language insights and recommendations.
-
Data Collection: TLOGic calculates operator performance metrics from your TLOG file (transaction counts, sales amounts, void/refund rates, timing metrics)
-
Prompt Creation: Your question or analysis request is combined with the operator data into a structured prompt
-
AI Processing: Google Gemini analyzes the data and generates insights based on patterns, outliers, and best practices
-
Results Display: The AI's analysis is displayed in an easy-to-read format with specific findings and recommendations
Accessing AI Analysis
- Navigate to the AI Playground tab after loading a TLOG file
- Review the operator data summary (automatically generated)
- Choose a pre-built prompt or write your own custom question
- Click "Analyze with AI"
- Wait for the analysis to complete (typically 5-15 seconds)
- Review the results and copy them if needed
Pre-Built Analysis Prompts
TLOGic includes four pre-built prompts designed for common analysis scenarios:
Top & Bottom Performers
"Analyze the operator performance data and identify the top 3 performers and bottom 3 performers. Explain what makes them stand out."
Best for: Identifying high performers for recognition and low performers who may need training or support. Provides specific metrics that differentiate top and bottom operators.
Flag Anomalies
"Identify any operators with unusually high void or return rates. What might be causing this?"
Best for: Detecting potential issues with accuracy, training, or possible policy violations. The AI considers context and suggests root causes.
Efficiency Analysis
"Compare the efficiency metrics (ring time, tender time) across operators. Who needs training?"
Best for: Identifying operators who may benefit from speed and efficiency training. Highlights slow transaction processing times.
Action Plan
"Provide 3 specific, actionable recommendations to improve overall operator performance based on this data."
Best for: Getting concrete next steps and improvement strategies based on the current team's performance patterns.
Writing Custom Prompts
For more specific analysis, write your own custom prompts in the text area. Here are tips for effective prompts:
Be Specific
Good Example:
"Which operators have void rates above 5% and what is their average transaction value compared to the team average?"
Avoid:
"Tell me about the operators."
Ask for Actionable Insights
Good Example:
"Based on the performance data, which 2 operators should I prioritize for advanced training and why?"
Avoid:
"Who is good?"
Combine Multiple Metrics
Good Example:
"Identify operators with both high sales amounts AND low void rates. What techniques might they be using that others could learn from?"
Common Use Cases
Training Needs
Prompt: "Which operators have ring times more than 20% slower than average? What specific areas should training focus on?"
Use this to identify skill gaps and create targeted training programs.
Recognition Programs
Prompt: "Rank operators by their composite performance score. Who are the top 5 and what specific achievements should be highlighted?"
Identify high performers for employee recognition and rewards.
Team Benchmarking
Prompt: "What is the median performance for each key metric? Which operators are above/below these benchmarks?"
Establish performance standards and identify outliers.
Loss Prevention
Prompt: "Are there any operators with unusual patterns in their void/refund activity that might warrant further investigation?"
Detect potential policy violations or fraud indicators.
Understanding AI Results
AI analysis results typically include:
- Specific Findings: Concrete observations about operator performance
- Comparative Analysis: How operators compare to each other and team averages
- Root Cause Suggestions: Possible reasons for performance patterns
- Recommendations: Actionable steps to improve performance
- Priority Guidance: Which issues or operators to address first
Pro Tip: Copy AI results using the "Copy" button and paste them into your performance review documents, training plans, or management reports. You can also run multiple analyses with different prompts to explore various angles of the same data.
Limitations & Considerations
Important Notes
- Data Quality: AI insights are only as good as the data in your TLOG file. Incomplete or inaccurate data will lead to less reliable analysis.
- Context Matters: The AI doesn't know your store's specific circumstances, policies, or staffing challenges. Use its insights as a starting point, not absolute truth.
- Privacy: Operator performance data is sent to Google Gemini for analysis. Don't use this feature if you have concerns about data privacy.
- Verification: Always verify AI findings with your own knowledge of the operators and business context before taking action.
Best Practices
Do This
- Run analysis after loading fresh, recent data
- Use specific, focused prompts
- Compare AI findings with your observations
- Run multiple analyses from different angles
- Save useful analyses for future reference
- Share insights with management teams
Avoid
- Making personnel decisions based solely on AI
- Running analyses on partial or corrupt files
- Using vague, open-ended prompts
- Ignoring context and circumstances
- Sharing raw AI output without interpretation
- Over-relying on AI instead of human judgment
Troubleshooting AI Analysis
AI Analysis Fails or Times Out
- Check your internet connection
- Ensure the TLOG file has operator data
- Try a shorter, simpler prompt
- Wait a moment and try again (temporary service issues)
Results Don't Make Sense
- Verify your TLOG file has complete, accurate data
- Check if the prompt was clear and specific
- Try rephrasing your question
- Run a pre-built prompt to verify AI is working correctly
Generic or Unhelpful Responses
- Make your prompt more specific
- Include metric thresholds (e.g., "above 10%")
- Ask for specific numbers and examples
- Request actionable recommendations