extractTextWithAI

This is an experimental feature powered by LLM technology. All feedback is welcome.

- extractTextWithAI: CAPTCHA value
- inputText: ${aiOutput}

Takes a screenshot and tries to extract text value from the screen using LLM. Output is then written into aiOutputvariable.

AI configuration

When to use?

extractTextWithAI could be a good fit when:

• View ID or content is not known beforehand

  • Search results

• Information on the screen is presented as an image

  • Promotional banners with embedded text

  • Captcha

extractTextWithAI is not intended to be a replacement for conventional element selectors (such as used with tapOn). When possible, prefer to use stable IDs or text values.

Examples

Amazon Search Results

Amazon reports its search results without any distinct IDs assigned to each item. We also don't know what will show up in the results beforehand. To work around this problem we can use AI to hint us what is the label of the first product item on this page and then tap on it:

- extractWithAI: Title of the first item on this page
- tapOn: ${aiOutput}

Configuration

AI configuration