> ## Documentation Index
> Fetch the complete documentation index at: https://algolia.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AI-assisted transformations

> How AI can help build no-code transformations.

export const Application = () => <Tooltip tip="An Algolia application is a self-contained environment with its own indices, configuration, and API keys. Applications don't share data or settings with each other.">
    application
  </Tooltip>;

Instead of using the form-based [no-code transformation builder](/doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/transform-your-data-without-code),
describe the transformation you need in plain English,
and AI will generate the configuration for you.
This is useful for complex conditions or computed values.

<img src="https://mintcdn.com/algolia/BaxS3Doxn613Z9Q0/doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_hero.png?fit=max&auto=format&n=BaxS3Doxn613Z9Q0&q=85&s=6315d8f8ee4fce159b24566ecd0c59da" alt="Screenshot of the AI transformation interface with a text box prompting the user to describe their desired change in plain English" width="1406" height="840" data-path="doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_hero.png" />

## When to use AI-Assisted transformations

This feature is for users who need to adjust data without writing code.
It helps when you want to:

* **Add new attributes.** For example, add a discount percentage field from the full price and the sale price.
* **Remove fields** that are no longer relevant.
* **Filter records.** For example, to exclude test products or filter by status.
* **Compute new values** from existing attributes. For example, create a ranking signal by combining popularity and newness.

### Key benefits

* **Simplifies complex logic.** You don't need to build nested "if/then" rules manually: AI handles it.
* **Faster iteration.** Quickly test and refine transformations using natural language.
* **You stay in control.** AI fills in the form, but you can review and adjust it before applying.
* **Preview before applying.** See how your data will look before committing changes.

## Use AI to build a transformation

1. Go to the [Algolia dashboard](https://dashboard.algolia.com/explorer/browse) and select your Algolia <Application />.

2. On the left sidebar, select **Data sources**.

3. Open the [**Connectors**](https://dashboard.algolia.com/connectors) page and select a connector.

4. Configure your data source by selecting the appropriate options.

5. Click **Create source**

6. Select **Create a new transformation** and then select **Transform without code**.

7. Click **AI Assist**

8. Describe your transformation by typing it in plain English. For example: "If price is greater than 100, add the label 'expensive'."

   <img src="https://mintcdn.com/algolia/BaxS3Doxn613Z9Q0/doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_english_request.png?fit=max&auto=format&n=BaxS3Doxn613Z9Q0&q=85&s=6bce3a671290d5f6ad1223fcb29f7c35" alt="Screenshot of the 'Delete Attribute: inventory' step with a text input and options to add conditions or groups." width="2894" height="1442" data-path="doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_english_request.png" />

9. **Review the suggestion.** AI generates a transformation configuration based on your prompt. You'll see a preview showing how your data will change before you apply it. Refine the transformation either by adjusting your prompt or by editing the form fields.

   <img src="https://mintcdn.com/algolia/BaxS3Doxn613Z9Q0/doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_preview.png?fit=max&auto=format&n=BaxS3Doxn613Z9Q0&q=85&s=ed2007586e9eed2d1a3754ca61d094c7" alt="Screenshot showing AI-generated transformation preview with editable field" width="2892" height="1432" data-path="doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_preview.png" />

10. Name the transformation and save it.

11. Complete the remaining connector workflow: select a destination, create the task that will schedule the job (and its indexing strategy), and then click **Run**. The connector then runs the transformation as part of the workflow.

    <img src="https://mintcdn.com/algolia/BaxS3Doxn613Z9Q0/doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_finish.png?fit=max&auto=format&n=BaxS3Doxn613Z9Q0&q=85&s=604c5c1eaa476c30d3b28d0288c07bac" alt="Screenshot of a task configuration interface with 'On demand' and 'Partial record updates' selected." width="1692" height="1408" data-path="doc/guides/sending-and-managing-data/send-and-update-your-data/how-to/ncdt_ai_finish.png" />

## Limitations

The AI assistant has a few constraints to keep in mind when using it with No Code data transformations.

* **Limited transformation types.**
  Only these transformation types are supported:
  *Add Attribute*, *Remove Attribute*, *Filter Records*, and *Compute New Value*. It doesn't cover advanced operations like joins, type conversions, or hierarchical restructuring.

* **Accuracy of suggestions.**
  The perfect transformation may not always be generated. Review and, if needed, refine the suggestion before applying it. This "human in the loop" step is essential.

* **Prompt constraints.**
  Prompts must be concise and specific. Long or ambiguous descriptions may lead to incomplete or incorrect transformations.

* **Performance limits.**
  Transformations are still bound by Algolia's [service limits](/doc/guides/scaling/algolia-service-limits) (for example, record size, memory, and execution timeout). AI can't bypass these constraints.

* **Security and data handling.**
  Only the record schema and your written prompt are sent to the AI model, not full record data. This means if you include sensitive data in a prompt, it *could* be processed, so avoid adding personally identifiable information or confidential values.

* **Over-reliance.**
  Use AI to help, not replace your judgment. Confirm that the previewed changes are correct, as incorrect logic could affect search relevance.

* **Language coverage.**
  Prompts are optimized for English. Support for other languages is limited.