Skip to main content
Enhancement lets you use AI to fill in, improve, or transform column values after extraction. This is useful for generating descriptions, translating content, or enriching data with web-sourced information.

​
How enhancement works

  1. Select rows and a column — Choose which rows to enhance and which column to fill.
  2. Write a prompt — Describe what the AI should generate, using context from other columns.
  3. AI processes each row — The model reads the row’s existing data and generates a value.
  4. Results are written back — Enhanced values replace or fill the target column.

​
Enhancement types

​
Content generation

Generate new content based on existing data: 
"Write a compelling product description based on the product name and specifications."

​
Translation

Translate content to another language, optionally using your glossary: 
"Translate the description to German."
Set target_language and use_glossary: true to apply your glossary terms during translation.

​
Web research

Enable web_search: true to let the AI search the web for additional information: 
"Find the current retail price for this product from major retailers."

​
Single column vs bulk

​
Tracking progress

Enhancement runs asynchronously. Check progress with: 
const status = await client.enhance.status("tbl_abc123");

for (const [column, stats] of Object.entries(status.columns)) {
  console.log(`${column}: ${stats.completed}/${stats.total} done`);
}

​
Credits

Each enhancement consumes credits. The estimated cost is returned when you start an enhancement job. Check your balance at any time via the Credits API.

​
Traces

Every AI-generated value has a trace showing the prompt, model used, confidence score, and sources. See Row traces for details.