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

# AI Field Fixer

> Use AI Field Fixer to clean and format extracted data, name fields from their content, or update field names and XPath expressions in batches with natural language.

AI Field Fixer lets you add, remove, or correct extraction fields with natural language. Describe how the fields or extracted values should change, review the proposed changes and sample values, then decide whether to apply them.

Introduced in Octoparse 10.2, AI Field Fixer is especially useful for cleaning and formatting extracted data, including tasks that would otherwise require selecting a formatting method or writing a regular expression. It can also interpret the meaning of already extracted content and rename multiple fields according to your requested naming convention, or update multiple XPath expressions in one request.

```text theme={null}
Describe the goal → review the proposed change → apply → test
```

## What AI Field Fixer does for you

Suppose the extracted values in a `Price` field include a currency symbol, such as `$123.00`, but you only want to keep the numbers. Enter:

> Remove the currency symbol from Price and keep only the number, for example 123.

AI Field Fixer then works through the change with you:

1. **Targets the existing field.** It connects your request to the current `Price` field.
2. **Creates the formatting rule.** It converts the request into an appropriate data-cleaning operation, using a regular expression when needed.
3. **Previews the output.** It applies the proposed rule to sample values so you can check the result.
4. **Accepts revisions.** If the output is not right, describe the adjustment in another sentence.
5. **Reviews batch changes together.** For multi-field requests, it presents the proposed updates together so you can confirm the names, formatting rules, or XPath expressions.
6. **Updates the task after approval.** The fields change only after you approve the proposal.

This removes the need to choose a formatting operation and write or test a regular expression manually.

## How much work it removes

AI Field Fixer reduces work in two areas: cleaning extracted data and updating multiple fields at once. When renaming fields, it can interpret the meaning of the extracted content and generate names that follow your requested naming convention.

| Task | Manual process | With AI Field Fixer |
| - | - | - |
| Clean or reformat a value | Choose a formatting method, write and test a regular expression, then inspect the preview | Describe the desired output in one sentence, review the generated result, and request another revision if needed |
| Remove unwanted text or symbols | Configure separate replacement or cleanup rules | Describe what to remove and review the cleaned sample values |
| Standardize dates or numbers | Select and configure the appropriate output format for each field | State the required format in natural language and preview the converted values |
| Refine an unsatisfactory result | Reopen the formatter and manually adjust its rules or regular expression | Describe the correction in another sentence and review the revised result |
| Rename multiple fields from their content | Inspect the values in each field, decide on a suitable name, and rename each field separately | Ask AI Field Fixer to interpret the extracted content and rename all fields according to your naming convention |
| Update multiple XPath expressions | Open each field and paste its XPath separately | Provide the target fields and known XPath expressions in one request, then review the proposed batch update |

AI Field Fixer is most effective when you already know how the extracted data should be cleaned or which field settings should change. It can understand the meaning of content that has already been extracted when generating field names, but it does not replace locating the correct webpage elements or testing the final task.

## Changes you can request

| Current field or value | Ask AI Field Fixer to... |
| - | - |
| `Price` contains `$123.00` | `Remove the currency symbol from Price and keep only the number.` |
| `Reviews` contains `1,234 reviews` | `Remove the word reviews and the comma; keep only the number.` |
| Dates use several formats | `Format all values in PublishedDate as YYYY-MM-DD.` |
| Several fields have unclear names | `Rename all fields based on their extracted content. Use PascalCase names, such as ProductName and CurrentPrice.` |
| Several fields have known replacement selectors | `Update ProductName to XPath A and Price to XPath B.` |

AI Field Fixer can clean and format extracted values and apply field-name or XPath changes in batches while leaving the surrounding workflow intact.

## Fix a field

<Steps>
  <Step title="Create an Auto-detected task">
    Run [Auto-detect](/docs/en/platform/auto-detect), then review the fields and sample values in the Data Preview.
  </Step>

  <Step title="Open AI Field Fixer from Data Preview">
    Generate the workflow and Data Preview first. Then open AI Field Fixer from the Data Preview header or the ✨ control beside a field. The entry remains unavailable until an extraction workflow and preview exist.
  </Step>

  <Step title="Describe the expected result">
    State how the existing value should be cleaned or formatted. For batch renaming, ask AI Field Fixer to infer suitable names from the extracted content and specify the naming convention you want. To update XPath expressions, list every target field and its replacement XPath explicitly or certain XPath format.
  </Step>

  <Step title="Select the element or provide an XPath">
    AI Field Fixer does not locate the target page element automatically. Select the intended element on the page or provide the XPath you want to apply.
  </Step>

  <Step title="Review the proposal">
    Compare the proposed change and sample values with the webpage. Check several rows, including items with different layouts or missing values.
  </Step>

  <Step title="Apply and test">
    Apply the change only when the preview is correct. Run a small test before starting a full local or cloud run.
  </Step>
</Steps>

## Where AI Field Fixer stops

AI Field Fixer in Octoparse 10.2 works with existing extraction fields. It can interpret already extracted content when generating field names, but it does not determine which webpage element represents the value you intend to collect, locate the correct XPath, or create or repair an entire workflow from a prompt.

Use the [no-code builder](/docs/en/platform/no-code-builder) to change pagination, scrolling, detail-page clicks, login and form actions, wait conditions, loops, or the overall action sequence.

<Warning>
  AI-generated changes can be incorrect. Always review the sample output and test the task before running it at scale.
</Warning>

## Related pages

<Columns cols={3}>
  <Card title="Auto-detect" href="/docs/en/platform/auto-detect">
    Generate the initial extraction fields and workflow.
  </Card>

  <Card title="No-code builder" href="/docs/en/platform/no-code-builder">
    Configure page actions and complex workflow behavior.
  </Card>

  <Card title="Refine data" href="/docs/en/platform/refine-data">
    Clean and reformat extracted values.
  </Card>
</Columns>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.