Privacy & Data Handling
Transparency on how data is handled in the TrainSight AI demonstration prototype.
Synthetic Demo Data
Default datasets consist solely of computer-generated mock entries. No real individuals or organizations are represented.
Session-Only Scope
Uploaded files are processed entirely in browser memory. There is no persistent database storing your evaluations.
Claude AI Processing
AI requests send aggregated statistical context to Anthropic only when you explicitly invoke AI features.
1. Demonstration Prototype Status
TrainSight AI is an experimental prototype designed to showcase modern AI-assisted analytics for organizational learning and talent teams. It is not an enterprise data processor, and users must not upload confidential, legally privileged, proprietary, or Personally Identifiable Information (PII).
2. Uploaded Dataset Handling
When you upload a CSV file via the Dataset Management interface:
- The file is parsed locally using client-side JavaScript (Papa Parse).
- Data is held in memory for the duration of your browser session.
- Refreshing the page or navigating away clears the uploaded dataset.
- No server database or external storage bucket retains your uploaded files.
3. Third-Party AI Services (Anthropic Claude)
When you trigger the “Generate AI Summary” button or ask questions in “Ask TrainSight AI”, a compact statistical summary and a limited subset of feedback comments are sent via server-side API route to the Anthropic Claude API for synthesis. No raw full-database dumps are transmitted.
4. User Responsibility
Please utilize only sanitized, public, or sample data when testing custom uploads. You are responsible for complying with your organization's data governance and confidentiality policies.