Integrate powerful language models from OpenAI, Anthropic, and Google into your workflows

AI Node

The AI node brings powerful language model capabilities into your workflows. It supports multiple AI providers and models, enabling text generation, analysis, code review, and intelligent data processing.

Configuration

Input Schema

Field	Type	Required	Description
`prompt`	`string`	Yes	The prompt to send to the AI model

Configuration Schema

Field	Type	Required	Default	Description
`provider`	`string`	No	`"anthropic"`	AI provider: "openai", "anthropic", or "google"
`model`	`string`	No	`"claude-4-sonnet-20250514"`	Model name to use
`api_key`	`string`	No	-	Custom API key (uses env vars by default)

Output Schema

Field	Type	Description
`output`	`JsonValue`	The AI model's response text

Basic Usage

Simple Text Generation

- id: generate_content
  type: ai
  input:
    prompt: "Write a brief summary of the benefits of workflow automation"
  configuration:
    provider: "anthropic"
    model: "claude-4-sonnet-20250514"

Code Analysis

- id: review_code
  type: ai
  input:
    prompt: |
      Review this code for potential issues:
      {{ code_snippet }}
      
      Provide specific feedback on:
      - Security vulnerabilities
      - Performance concerns
      - Best practices
  configuration:
    provider: "openai"
    model: "gpt-4"

Advanced Examples

- id: analyze_data
  type: ai
  input:
    prompt: |
      Analyze this sales data and provide insights:
      {{ sales_data.output }}
      
      Please include:
      - Key trends and patterns
      - Anomalies or outliers
      - Actionable recommendations
      
      Format the response as structured JSON.

- id: validate_analysis
  type: ai
  input:
    prompt: |
      Validate this analysis for accuracy and completeness:
      {{ analyze_data.output }}
      
      Are the conclusions supported by the data?

Perfect for business intelligence, financial analysis, and data-driven decision making workflows.

- id: gather_context
  type: search
  input:
    query: "{{ topic }}"
    maxResults: 5

- id: generate_article
  type: ai
  input:
    prompt: |
      Based on this research context:
      {{ gather_context.output }}
      
      Write a comprehensive article about {{ topic }}.
      Include relevant facts, statistics, and insights.
      Target audience: {{ target_audience }}
  configuration:
    provider: "anthropic"
    model: "claude-4-sonnet-20250514"

Ideal for marketing content, research reports, and educational material creation with fact-based foundations.

- id: evaluate_options
  type: ai
  input:
    prompt: |
      Evaluate these options for {{ decision_context }}:
      
      Options:
      {{ options }}
      
      Criteria:
      {{ evaluation_criteria }}
      
      Provide a structured recommendation with:
      - Pros and cons for each option
      - Risk assessment
      - Final recommendation with reasoning
      
      Response format: JSON

Essential for strategic planning, vendor selection, and complex business decisions requiring structured analysis.

Tool Integration

The AI node includes access to a common toolset for enhanced capabilities:

- id: ai_with_tools
  type: ai
  input:
    prompt: |
      Help me research {{ topic }} and create a summary.
      Use available tools to gather current information.

Tools automatically available:

Web search capabilities
Data processing functions
Structured output generation

Best Practices

Prompt Engineering

Clear Instructions:

input:
  prompt: |
    Task: Analyze customer feedback
    Input: {{ feedback_data }}
    Output format: JSON with sentiment, themes, and action items
    Context: SaaS product feedback analysis

Structured Prompts:

input:
  prompt: |
    ## Context
    {{ context }}
    
    ## Task
    {{ specific_task }}
    
    ## Requirements
    - {{ requirement_1 }}
    - {{ requirement_2 }}
    
    ## Output Format
    {{ desired_format }}

Error Handling

- id: ai_task
  type: ai
  input:
    prompt: "{{ complex_prompt }}"
  configuration:
    errorHandling: "continue"
    retryPolicy:
      maxRetries: 2

- id: validate_output
  type: condition
  input:
    if: "{{ ai_task.output && ai_task.output.length > 0 }}"
    then: ["process_result"]
    else: ["handle_ai_error"]

Cost Optimization

# Use appropriate models for task complexity
- id: simple_classification
  type: ai
  configuration:
    provider: "openai"
    model: "gpt-4o-mini"  # Faster, cheaper for simple tasks

- id: complex_analysis
  type: ai
  configuration:
    provider: "anthropic"
    model: "claude-4-sonnet-20250514"  # More capable for complex tasks

AI Node

On this page