AI Field Fixer is a new feature in Octoparse 10.2 that lets you update data fields through a conversation with AI. Instead of editing fields one by one, building a series of Clean Data steps, or writing regular expressions for every field, you can describe the result you want in plain language.
You can use AI Field Fixer to:
Update the XPaths of multiple fields in one request
Reformat extracted values by describing the desired output
Rename multiple fields based on their sample content
Availability: AI Field Fixer is available in Octoparse 10.2.
How to use AI Field Fixer
Open your task and go to Data Preview.
Open AI Field Fixer.
Describe the changes you want in the chat box. You can include the target format, naming convention, XPath pattern, and any exceptions the AI should keep.
Review the suggested changes, continue the conversation if anything needs adjusting, and confirm the replacement when it looks correct.
You do not need to fit every requirement into your first message. If the first result is not exactly what you need, reply with a more specific instruction, such as “keep the currency code" or or “use snake_case."
Use case 1: Rename fields based on their content
Auto-detected fields or manually created fields may contain generic names such as Field_1, Text_2, and Field_3. AI Field Fixer can examine the sample content and propose meaningful names for multiple fields at once.
Example prompt:
Rename the fields based on their sample content. Use short, descriptive English names in snake_case. Keep the current field order, values, and XPaths unchanged.
For example, the AI may identify values such as $49.99, In stock, and 4.8 and suggest names such as price, availability, and rating.
You can also provide your own naming rules:
Rename the fields based on their content. Use PascalCase and add the prefix “Product” to every name. Do not change any field values or XPaths.
Use case 2: Reformat field values without Clean Data or regular expressions
Many extraction tasks need consistent output: prices without currency symbols, dates in one standard format, or text without extra labels and spaces. Previously, you might have added multiple Clean Data steps or written a regular expression. AI Field Fixer lets you describe the desired result instead.
Example: Standardize prices
If the source values look like $1,299.00 USD and you need 1299.00, try:
Format the Price field as a number with two decimal places. Remove currency symbols, currency codes, and thousand separators. Leave missing values blank.
Example: Standardize dates
If the website uses values such as Sep 2, 2026, try:
Convert the Date field to YYYY-MM-DD. Leave values blank when a valid date cannot be identified.
Example: Remove labels and extra spaces
For the SKU field, remove the “SKU:” label, trim leading and trailing spaces, and keep the remaining letters, numbers, and hyphens unchanged.
A strong formatting request normally includes four things: the fields to change, the desired output, what to remove or preserve, and how to handle blank or invalid values.
Use case 3: Update multiple field XPaths in the same way
When a website changes its page structure, several fields may stop returning data at the same time. Editing each XPath separately can be slow, especially when the fields should follow the same pattern. With AI Field Fixer, you can offer the field XPath to AI and let it to apply for the same XPath pattern to multiple fields.
Example: Apply a specific XPath pattern
If you already know the structure you want, include it directly in the request.
Example prompt:
Update the selected product-link fields to use an XPath in this format: .//a[contains(@class, 'product-link')]/@href.
Tip: Tell the AI whether the fields are inside a loop and whether you want relative or absolute XPaths. After applying the changes, check several list items and detail pages to make sure each field still captures the intended element.
How to write better prompts
Define the scope. Say whether the request applies to all selected fields or only specific fields.
Show the target format. Include an example such as
2026-09-02,1299.00, orproduct_price.State what must not change. For example, ask the AI to preserve field order, values, names, or XPaths.
Explain exceptions. Tell the AI how to handle blank values, invalid dates, unexpected text, or fields with unclear content.
Refine the result conversationally. Add a follow-up instruction instead of starting over when only part of the result needs adjustment.
Review before running your task
AI Field Fixer can save significant setup time, but the suggested changes should still be reviewed. Test the updated fields against different pages and edge cases, especially when a site has dynamic layouts, optional fields, or inconsistent value formats.
If a result is not correct, return to the conversation and describe the adjustment you need. The more specific your examples and constraints are, the more predictable the result will be.