Key Value Storage Blocks

What is key-value storage?

• Think of key-value storage as a digital filing cabinet where you store information under specific labels. Each piece of information (the "value") is stored with a unique label (the "key") that allows you to easily find and retrieve it later.
  • In everyday terms:
    • A key is like a name tag or label
    • A value is the information you want to save
      • Together, they form a "key-value pair"
      • For example, if you want to save someone's favorite color, the key might be "JohnsFavoriteColor" and the value would be "blue".

Why use key-value storage in your workflows?

• Store information between different steps of your workflow
• Remember information between different workflow runs
• Create simple counters or flags
• Save user preferences or settings
• Track the state of multi-step processes

The Three Key-Value Storage Blocks in Scout

Scout offers three simple blocks for working with key-value storage:

1. Key-Value Storage (Write) - Saves information
2. Key-Value Storage (Read) - Retrieves saved information
3. Key-Value Storage (Delete) - Removes saved information

Let's explore each one in detail!

1. The Key-Value Storage (Write) Block

• What it does

  • This block allows you to save information that you might need later in your workflow or in future workflow runs.

• How to use it

  • You need to provide two things:
    • Key - A unique label for your information (required)
    • Value - The actual information you want to store (required)

• Practical Examples

  • Example 1: Storing Basic User Information

    • Store a customer's name for personalization:
      • Key: customer_name
      • Value: Maria Garcia

  • Example 2: Using Dynamic Values from Workflow Input

    • Saving user input from a form:
      • Key: feedback_{{inputs.ticket_id}}
      • Value: {{inputs.user_feedback}}
        • This creates a unique key for each feedback submission based on the ticket ID.

  • Example 3: Storing JSON Data

    • Save structured information:
      • Key: user_profile_1234
      • Value: {"name": "John Smith", "email": "john@example.com", "preferences": {"theme": "dark", "notifications": true}}

  • Example 4: Storing Results from Previous Blocks

    • Save the output from an AI model for later use:
      • Key: summarization_{{workflow_run_id}}
      • Value: {{llm_summarize.output}}

Best Practices for Writing Data

• Use descriptive key names that clearly indicate what data is stored
• Add prefixes to keys for organization (e.g., user_, session_, temp_)
• Include identifiers in keys when storing multiple related items (e.g., user_123_preferences)

2. The Key-Value Storage (Read) Block

• What it does

  • This block retrieves information you've previously saved using the Write block.

• How to use it

  • You need to provide:
    • Key - The exact label you used when saving the information (required)
    • Output As - The format you want the retrieved information in (optional, defaults to text)

• Output Format Options

  • text - Plain text (default)
  • number - Numeric value
  • bool - True/False value
  • json - JSON data (for objects and lists)
    

• Practical Examples

  • Example 1: Using the Retrieved Value in a Message

    • After retrieving the customer name, you can use it in a message template:
      • Hello {{customer_name_read.output}}! Welcome back to our service.
      • customer_name_read is the ID of your Read block.

  • Example 2: Reading Dynamic Keys Based on Current Input

    • Retrieve specific feedback using a ticket ID:
      • Key: feedback_{{inputs.ticket_id}}
      • Output As: text
        • This dynamically constructs the key based on the current ticket ID being processed.

    • Example 3: Reading JSON Data

      • Retrieve a stored user profile:
        • Key: user_profile_1234
        • Output As: json
          • This retrieves the stored JSON object so you can access individual properties like {{user_profile_read.output.name}} or {{user_profile_read.output.name.company}} in subsequent blocks.

  • Common Use Cases

    • Personalization: Retrieve user preferences to customize responses
    • Session Management: Read session data to maintain context across interactions
    • Progress Tracking: Check completion status of multi-step processes
    • Configuration: Access stored settings or parameters
    • Data Persistence: Retrieve previously stored calculation results

3. The Key-Value Storage (Delete) Block

• What it does

  • This block removes information you no longer need from the storage, helping keep your storage clean and organized.

• How to use it

  • You only need to provide:
    • Key - The label of the information you want to delete (required)

• Practical Example

  • Example 1: Cleanup After Workflow Completion

    • Remove all processing data once a workflow completes:
      • Key: processing_{{inputs.workflow_run_id}}

• When to Use the Delete Block

  • Privacy concerns: Remove sensitive user data when no longer needed
  • Housekeeping: Clean up temporary values to avoid cluttering your storage
  • Reset state: Clear previous states before starting a new process
  • Error recovery: Remove invalid or corrupted data
  • Compliance: Follow data retention policies by removing outdated information

Micro-Challenge:

• To practice working with key-value blocks, start by using the Key-Value Playground template to explore how they function. As an additional exercise, try adding a delete block to the workflow to see how it removes data from the key-value store.