📖 Description
🔹 How it works
This workflow introduces an AI + Human-in-the-Loop pipeline for employee timesheet management. It combines the power of Google Drive, AI (OCR + LLM), and Gmail with a human review step to ensure accuracy and compliance.
-
AI-Powered File Discovery
- Scans a Google Drive folder for new or updated timesheet files (PDF, Word, Excel, Images).
-
AI Data Extraction
- Uses OCR and LLM (Mistral) to intelligently read and extract structured data.
- Supports multiple formats: PDF, Word (DOC/DOCX), Excel (XLS/XLSX), and Image files (JPG, PNG, scanned documents).
- Creates clean JSON with file details and timesheet logs (date, hours worked, tasks, notes).
-
Smart Data Formatting
- Converts AI output into a clear HTML summary table for easy review.
- Flags potential anomalies (missing hours, duplicate dates, irregular entries).
-
Human-in-the-Loop Verification
- Sends an approval email via Gmail containing:
- File metadata
- AI-generated HTML summary
- JSON attachment of raw extracted data
- HR/Managers review the summary and approve/reject before final actions occur.
-
Post-Approval Automation (optional)
- Approved records can be saved in a separate Google Drive folder.
- Employees or HR receive confirmation emails.
⚙️ Set up steps
-
Connect Credentials
- Add Google Drive and Gmail credentials in n8n.
- Configure Mistral (or any LLM) API credentials.
-
Configure Google Drive
- In the “Search files and folders” node, replace the
folderId
with your company’s timesheet folder ID.
-
Customize Extraction Schema
- Sticky notes explain how JSON output is structured.
- Adapt it for your organization’s needs (e.g., overtime, project codes).
-
Set Up Human Verification Emails
- Update Gmail node recipients to your HR or approval team.
- Customize the email body (AI summary + JSON file attached).
-
Activate & Test
- Enable the workflow.
- Upload a sample timesheet to trigger the AI + human verification loop.
⚡ Result: A robust AI + Human-in-the-Loop workflow that reduces repetitive data entry, prevents payroll errors, and gives HR full confidence before final approval.