Section 1 of 1
Industrial-Scale Content Generation
Batch processing is where PromptForge Lab truly shines. Instead of running one prompt at a time, you can process hundreds or thousands of data rows through the same prompt template simultaneously.
Prepare Your CSV
Each column becomes a variable. Column header 'customer_name' maps to {{customer_name}} in your prompt. Ensure clean data — no special characters in headers.
Map Variables
The Lab auto-detects your CSV columns and highlights matching {{variables}} in your prompt. Unmatched variables are flagged in red.
Set Concurrency
Choose how many rows to process simultaneously. Higher concurrency = faster, but may hit API rate limits. Start with 5, scale to 20.
Run & Monitor
Watch real-time progress. Each row shows: Status (⏳ Processing, ✅ Complete, ❌ Error), Token Count, and Latency.
Export Results
Download the complete results as a new CSV with all original columns plus an 'ai_output' column. Or export as individual files.
Start with a 5-row test run before processing the full batch.
Include a 'row_id' column in your CSV for easy tracking and debugging.
Use the Lab's 'Retry Failed' button to re-run only the rows that errored.
Pre-validate your CSV data — garbage in = garbage out.
Don't upload CSVs with more than 10,000 rows without testing a sample first.
Don't use commas inside data fields without proper CSV quoting.
Don't set concurrency to maximum on a shared API key — you'll hit rate limits.