AI Actions

Introduction

AI Actions in Locale.ai enable users to leverage Large Language Models (LLMs) within their workflows, enhancing automation capabilities with artificial intelligence. These actions allow for sophisticated text generation, classification, and data extraction tasks, making workflows more intelligent and adaptive.

Available LLM Models

Locale currently offers users a choice between two powerful LLM models:

  1. Claude Sonnet
  2. Chat GPT

Users can select their preferred model when configuring AI actions in their workflows.

Types of AI Actions

Locale provides three distinct AI actions that users can incorporate into their workflows:

1. AI Text Generator

The AI Text Generator action allows users to create dynamic, context-aware text based on a given prompt.

How it works:

  • Users provide a prompt for the AI to generate text.
  • Data variables can be included within the prompt to offer additional context.
  • The AI generates text based on the prompt and any provided context.

Example Use Case:

Generating personalized email responses based on customer inquiries.

Prompt: Write a friendly response to a customer inquiry about {{product_name}}. 
The customer's name is {{customer_name}} and their question was: "{{customer_question}}".

2. AI Classifier

The AI Classifier action categorizes input text into predefined classes based on a given prompt and classification parameters.

How it works:

  • Users provide a prompt with context for classification.
  • Classification parameters are defined by the user, each with a unique key and description.
  • The AI classifies the input into one of the given parameters.

Example Use Case:

Categorizing customer feedback into sentiment categories.

Prompt: Classify the following customer feedback into one of the provided categories.

Feedback: {{customer_feedback}}

Classification Parameters:
- positive: The feedback expresses satisfaction or approval.
- neutral: The feedback is neither clearly positive nor negative.
- negative: The feedback expresses dissatisfaction or criticism.

3. AI Data Extractor

The AI Data Extractor action extracts specific information from given text based on user-defined parameters.

How it works:

  • Users provide a prompt with context for data extraction.
  • Extraction parameters are defined by the user, each with a unique key and description.
  • The AI extracts the requested information based on the provided parameters.

Example Use Case:

Extracting key details from customer support call transcripts.

Prompt: Extract the following information from this customer support call transcript:

Transcript: {{call_transcript}}

Extraction Parameters:
- customer_name: The name of the customer
- product_mentioned: Any product names mentioned in the call
- main_issue: The primary problem or concern raised by the customer
- resolution_status: Whether the issue was resolved (resolved/unresolved)

Configuring AI Actions in Workflows

To use AI actions in your Locale workflows:

  1. Add the desired AI action node to your workflow.
  2. Choose between Claude Sonnet or Chat GPT as the LLM model.
  3. Configure the action-specific settings (prompt, classification parameters, or extraction parameters).
  4. Use data variables to incorporate dynamic content from previous nodes in the workflow.

Best Practices

  • Provide clear and specific prompts to get the most accurate results.
  • Use data variables to make your AI actions dynamic and context-aware.
  • Test your AI actions with various inputs to ensure they perform as expected across different scenarios.
  • For classification and data extraction, provide detailed descriptions for each parameter to guide the AI's decision-making process.

AI Actions in Locale bring the power of advanced language models to your workflows, enabling more intelligent automation, personalized content generation, and sophisticated data processing. By leveraging these AI capabilities, users can create more efficient, adaptive, and powerful workflows that can handle complex tasks with ease.

Table of Contents