Advanced AI Nodes, Conditional Decision Logic and Dynamic Routing

Discover how to branch execution with conditional logic (IF / ELSE), deploy generative AI nodes powered by Knowledge Base RAG, and route leads based on semantic intent classification.

⚡Mandatory Prerequisites & Initial Checklist

✓Proficiency with basic canvas workflow authoring.
✓Company FAQs and reference files uploaded to the Knowledge Base module.

📋Step-by-Step Instructions in Dashboard

Implement Conditional Decision Node (IF / ELSE Branching)

👉 Where to Click:Automation Canvas > Logic Nodes > 'Conditional Branch'

Conditional nodes split flows into two distinct branches ('True' and 'False') based on defined parameters:

Field Dictionary & Parameters:

Condição por TagVerifica se o contato possui ou não possui uma tag específica (ex: Se tiver a tag 'Cliente Ativo', segue caminho A; senão, caminho B).

Condição por Valor de CampoCompara valores numéricos ou texto (ex: Se 'Faturamento' for maior que R$ 50.000, manda para o vendedor sênior).

Condição por Horário de ExpedienteCheca se o contato está chegando dentro do horário comercial da empresa (ex: 08h às 18h). Se estiver fora, envia mensagem informando que a equipe responderá pela manhã.

Add Generative AI Response Node with RAG Knowledge Retrieval

👉 Where to Click:Canvas > AI Nodes > 'Smart AI Response (RAG)'

Instead of hardcoding rigid robot replies, drop an AI Response node. Configure persona directives (e.g. 'You are Sophia, senior consultant, helpful and professional') and assign Knowledge Base folders. When prospects ask technical or pricing questions, AI retrieves facts and drafts natural, accurate replies in seconds.

Deploy Semantic AI Intent Router Node

👉 Where to Click:Canvas > AI Nodes > 'AI Intent Router'

The AI Intent Router classifies lead messages into multiple output branches without relying on exact keywords. E.g.: 1) 'Purchase Intent' -> create CRM deal and alert closer; 2) 'Support / Tech Question' -> route to helpdesk; 3) 'Complaint / Churn' -> route to supervisor. Regardless of how the prospect phrases their message, semantic AI ensures correct dispatch.

🚨Critical Safety Rules & Ban Protection

🛡️Hallucination Safeguard: In AI prompt settings, keep active the guardrail directive: 'If the answer is not contained in reference files, politely offer to transfer to a human specialist'.
🛡️Token Budget Protection: Set maximum token limits per response (recommended: 250 tokens) to prevent excessively verbose messages on WhatsApp and regulate AI credit burn.

❓Frequently Asked Questions & Troubleshooting

P:What happens if the AI fails to classify intent?

The node includes an 'Unclassified / Fallback' exit port. Connect it to an alternative prompt asking the user to clarify or transfer to live support.

P:Can the AI deliver PDF files or images automatically?

Yes! You can configure the AI to attach your pricing sheet PDF or product catalog whenever requested by the prospect.

← Visual Automations•← Back to Help Center

MailerWeb Knowledge Base • Atualizado em 2026 • Documentação Oficial do Usuário