Skip to main content

Overview

AI nodes bring the power of large language models (LLMs) and generative AI to your workflows. Generate text, analyze content, create images, and build intelligent automations.

Available AI nodes

When to use AI nodes

Use caseRecommended node
Content generationLLM
Content analysisLLM
Complex reasoningAI Agent
Tool-assisted tasksAI Agent
Data analysisData Analyzer
Image understandingImage to Text
Image creationText to Image

Model selection

Different models have different strengths:
ModelBest forSpeedCost
GPT-4Complex reasoning, long contentMediumHigh
GPT-3.5Simple tasks, high volumeFastLow
ClaudeLong documents, nuanced analysisMediumMedium
GeminiMultimodal, Google integrationFastMedium

AI settings explained

Temperature

Controls creativity vs. consistency:
ValueBehaviorUse for
0.0-0.3Very consistent, focusedData extraction, factual
0.4-0.6BalancedGeneral use
0.7-1.0Creative, variedCreative writing, brainstorming

Max tokens

Limits response length:
  • 500 tokens ≈ 375 words
  • 1000 tokens ≈ 750 words
  • 4000 tokens ≈ 3000 words
Set max tokens based on your expected output length. Higher values cost more.

Top P / Top K

Fine-tune output diversity:
  • Top P (0-1): Probability mass for token selection
  • Top K (1-100): Number of tokens to consider
For most use cases, leave at defaults.

Best practices

Write clear prompts

Good prompts lead to good outputs:
❌ "Summarize this"

✅ "Summarize this article in 3 bullet points, 
   focusing on key takeaways for marketers. 
   Keep each bullet under 20 words."

Use system messages

Set context for the AI:
System: You are a senior SEO consultant analyzing 
website content. Be specific and actionable.

User: Analyze this page for SEO issues...

Structure your output

Request specific formats:
Output as JSON:
{
  "summary": "...",
  "score": 0-100,
  "recommendations": ["...", "..."]
}

Handle errors gracefully

AI outputs can be unpredictable:
  • Validate JSON outputs
  • Set reasonable max tokens
  • Add retry logic for failures

Common patterns

Content generation pipeline

Analysis and extraction

Intelligent routing

Next steps