AI Nodes

Add AI-generated responses to any workflow. Nodebase includes 7 AI provider nodes — OpenAI, Anthropic, Gemini, Groq, DeepSeek, Perplexity, and xAI. All share the same interface: a prompt, an optional system prompt, and an output with the response text and token usage.

Provider Comparison

NodeDefault ModelBest ForVariable Prefix
OpenAIgpt-4oGeneral purpose, function calling, vision{{openAI.content}}
Anthropicclaude-sonnet-4-5Long context, complex reasoning, document analysis{{anthropic.content}}
Geminigemini-2.0-flashFast, multimodal (image + text){{gemini.content}}
Groqllama-3.3-70bFastest inference — ideal for real-time chatbots{{groq.content}}
DeepSeekdeepseek-chatCost-effective, strong at coding tasks{{deepSeek.content}}
Perplexitysonar-proWeb search included — answers with citations{{perplexity.content}}
xAIgrok-2-latestReal-time knowledge, current events{{xAI.content}}

Common Configuration

All AI nodes share the same fields:

FieldRequiredDescriptionSupports Variables
CredentialYesAPI key for the selected providerNo
ModelYesModel name — see provider comparison table aboveNo
PromptYesThe user message or question to send to the AIYes
System PromptNoInstructions that set the AI's behaviour and personaYes

Output Variables

All AI nodes output the same structure (replace nodeName with the variable name configured for that node):

jsonOutput
{
  "nodeName": {
    "content": "Your order #SHP-001 has been dispatched and will arrive in 3-5 days.",
    "model": "gpt-4o",
    "usage": {
      "prompt_tokens": 85,
      "completion_tokens": 24,
      "total_tokens": 109
    }
  }
}

Setting Up Credentials

ProviderAPI Key Location
OpenAIplatform.openai.com → API keys
Anthropicconsole.anthropic.com → API Keys
Geminiaistudio.google.com → Get API Key
Groqconsole.groq.com → API Keys
DeepSeekplatform.deepseek.com → API Keys
Perplexitywww.perplexity.ai → Settings → API
xAIconsole.x.ai → API Keys
  1. Go to Settings → Credentials
  2. Click Add Credential
  3. Select the provider (e.g. OpenAI)
  4. Paste the API key
  5. Click Save

Complete Workflow Examples

WhatsApp AI Customer Support Bot

Use case: Automatically answer customer questions on WhatsApp using Groq for fast responses.

textWorkflow
WhatsApp Trigger (messageTypes: text)
→ Groq
    model:        llama-3.3-70b
    systemPrompt: "You are a helpful customer support agent for Priya's Clothing Store.
                   Answer questions about orders, returns, and products.
                   Be concise — max 3 sentences."
    prompt:       "Customer message: {{whatsappTrigger.text}}
                   Customer name: {{whatsappTrigger.senderName}}"
→ WhatsApp — Send Message
    to:      {{whatsappTrigger.from}}
    message: {{groq.content}}

AI Order Confirmation Email

Use case: Generate a personalised, friendly order confirmation email using OpenAI.

textWorkflow
Razorpay Trigger (payment.captured)
→ OpenAI
    model:        gpt-4o-mini
    systemPrompt: "You are a friendly email writer for an Indian D2C clothing brand.
                   Write warm, professional emails in English."
    prompt:       "Write an order confirmation email for:
                   Customer: {{razorpayTrigger.payload.payment.entity.notes.name}}
                   Amount: ₹{{razorpayTrigger.payload.payment.entity.amount}} (note: divide by 100)
                   Payment ID: {{razorpayTrigger.payload.payment.entity.id}}
                   Keep it under 100 words."
→ Gmail — Send Email
    to:      {{razorpayTrigger.payload.payment.entity.email}}
    subject: "Your order is confirmed! 🎉"
    body:    {{openAI.content}}

Market Research with Perplexity

Use case: Get up-to-date market data for a product category with web search included.

textWorkflow
Webhook Trigger (POST /research, body: { category })
→ Perplexity
    model:  sonar-pro
    prompt: "What are the top 5 trending products in the {{body.category}} category
             in India this month? Include prices and where to buy."
→ Slack — Send Message
    channel: #market-research
    text:    "Research for {{body.category}}:
{{perplexity.content}}"

Choosing the Right Provider

Use CaseRecommended ProviderReason
Real-time WhatsApp chatbotGroqFastest inference — sub-second responses
Long document analysisAnthropic200K context window, best at following instructions
Cost-sensitive bulk processingDeepSeekCheapest per token, good quality
Research with current dataPerplexityBuilt-in web search — no need for separate HTTP request
Image + text tasksGeminiMultimodal support with fast flash model
General purpose / function callingOpenAIBest ecosystem, most reliable, strong reasoning
Current events / newsxAIReal-time knowledge cutoff

Common Issues & Solutions

IssueCauseSolution
401 UnauthorizedInvalid or expired API keyRe-generate the API key from the provider dashboard and update the credential
429 Rate LimitToo many requests to the providerUpgrade your provider plan or add a Wait node between AI calls
Response is cut offMax tokens limit reachedIncrease the max_tokens setting or shorten your prompt
Slow response in WhatsApp botUsing a large modelSwitch to Groq (llama-3.3-70b) for fastest response time

Related Nodes

  • WhatsApp Trigger — use AI to respond to incoming messages
  • Gmail — send AI-generated email content
  • Code — post-process AI output with custom JavaScript
  • Set Variable — save AI response to a named variable for reuse
RazorpayWhatsApp