The Filter results block acts as an intelligent gatekeeper in your workflow. It allows you to inspect the data or content provided by a source block (like Open datasets or Read RSS feeds) and decide whether the workflow should proceed to the data extraction phase for a specific item. This is essential for creating efficient and focused automations. By filtering upfront, you ensure that resource-intensive extraction blocks like Find Links or Extract Data only run on relevant items.

​
Purpose

Use the Filter results block before an extraction step to:
  • Process only the items from a data source that contain specific keywords.
  • Filter content based on a date (e.g., only process articles published after a certain date).
  • Use AI to apply a complex, descriptive filter to text (e.g., only pass through items with a “positive sentiment”).
  • Prevent the workflow from running on irrelevant or unimportant data, saving time and resources.

​
Configuration

The block’s configuration is centered around a dropdown menu that lets you choose the filtering method. After you select an option that requires input, a text field will appear to specify the condition.
  1. Dropdown Menu Options:
    • No filter, pass through all changes: This is the default state. The block will not perform any filtering, and all items from the previous step will be passed through to the next.
    • Look for specific keywords (one per line): After selecting this, a text field appears. Provide a list of keywords, one per line. The workflow will only continue for an item if its content contains at least one of the specified keywords.
    • Describe the filter in English, and filter with AI: After selecting this, a text field appears. Provide a natural language instruction describing your filtering criteria. The AI will analyze the content of each item and determine if it matches your description.
      • Example condition: The text should be a positive customer review or The article must be about financial technology (fintech).
    • The content must mention a specific date or time which must be after…: After selecting this, a date/time picker appears. The workflow will only continue for items whose content contains a date that is more recent than the one you specified.
[Screenshot: Filter results block configuration panel]

​
How It Works

  1. The Filter results block is placed after a block that produces a list of items (e.g., Read RSS feeds outputs a list of posts, Open datasets outputs a list of rows).
  2. The workflow iterates through each item from the preceding block.
  3. For each item, the Filter results block applies your chosen filter rule.
  4. If the item passes the filter, the workflow continues to the subsequent blocks for that item.
  5. If the item does not pass the filter, the workflow stops for that specific item and moves on to the next one in the list.

​
Example: Processing Only Relevant News Articles

Imagine a workflow that monitors a news RSS feed but should only analyze articles about a specific company.
  1. Read RSS feeds Block:
    • Configured to fetch the latest 10 articles from a news source.
  2. Filter results Block (added next):
    • Filter method chosen: Look for specific keywords (one per line)
    • Add the filter condition here: 
      Jsonify
OpenAI
  3. Extract Data Block (added next):
    • Configured to summarize the article content.
  • Result: The workflow will check all 10 articles fetched from the RSS feed. If an article’s title or description contains either “Jsonify” or “OpenAI”, it will pass the filter, and the Extract Data block will run on it to create a summary. Articles that do not contain these keywords will be ignored.

​
Key Considerations

  • Placement is Crucial: This block must be placed after a block that provides iterable data (a list of items).
  • AI vs. Keywords: For simple text matching, the keyword filter is fast and precise. For more nuanced filtering based on meaning, context, or sentiment, the descriptive AI filter is much more powerful.
  • Efficiency: Using a filter early in your workflow is a key strategy for efficiency, as it prevents resource-intensive blocks (like complex Extract Data or LLM Completion steps) from running on irrelevant data.
The Filter results block allows you to add powerful conditional logic to your automations, making them smarter and more efficient.