Bulk Processing
Bulk Processing lets you run an AI operation over an entire CSV file — row by row — at scale. Instead of copying and pasting data into a chat window one entry at a time, you upload the whole dataset, configure the operation once, and let Hubrix process every row in parallel.
What you can do with Bulk Processing
| Use case | Operation type |
|---|---|
| Score or label support tickets by sentiment | Classify |
| Extract invoice fields (vendor, date, amount) from text | Extract |
| Write a product description for each SKU | Prompt |
| Translate customer reviews to English | Translate |
| Analyse each row through your Data Analyst agent | Agent |
How it works
Bulk Processing uses a 3-step wizard:
- Choose operation — select one of the five operation types (Prompt, Agent, Classify, Extract, or Translate).
- Upload CSV — upload your data file (UTF-8, header row required, max 10,000 rows).
- Configure — map CSV columns to the operation's input fields, adjust settings like parallelism.
Once configured, click Run and Hubrix starts processing rows. A live progress bar and SSE-powered status feed show results in real time.
Parallelism
The parallelism slider controls how many rows are processed simultaneously. Higher parallelism finishes jobs faster but consumes credits faster. Lower parallelism is gentler on your credit balance and less likely to hit rate limits on underlying AI models.
Parallel processing means rows are not necessarily completed in the order they appear in your CSV. The output preserves the original row order regardless.
Viewing results
The Results tab shows all rows with their output columns. Filter by status (completed, failed, pending) to focus on issues.
The Charts tab shows analytics for completed jobs — status breakdowns, categorical distributions, numeric histograms.
Exporting
Download results as CSV (all rows plus output columns) or XLSX (formatted spreadsheet) from the Download button once the job completes or partially completes.
What's next
- CSV Upload — formatting requirements and column mapping
- Operations — what each of the five operation types does
- Charts — analytics on your completed job results
- Export — download CSV or XLSX results
Was this helpful?