Why does my AI prompt accuracy decrease when processing large batches (400+ companies) compared to small batches (1-10 companies)?

Large batch processing can reduce AI prompt accuracy due to rate limiting, system stability measures, and potential context window overload when many requests are processed simultaneously. Clay implements rate limits per row/column to ensure system stability, which can impact performance on large batches. To maintain accuracy while processing large datasets: (

1. Use the "Choose number of rows to run" feature in the AI column header to process 10-25 rows at a time, (

2. Test different batch sizes to find your accuracy sweet spot, (

3. Process during off-peak hours for better performance, and (

4. Use "Force" rerun or "Troubleshoot with AI" on questionable results to verify if issues were due to rate limiting. Each AI instance processes rows independently, so smaller batches help preserve prompt adherence and accuracy.