Debate and validate chat answers with GPT‑5, Claude Sonnet and Claude Opus
Ready for 3.0What this workflow does
This AI workflow has 19 nodes and runs when a chat message arrives. Along the way it connects OpenAI Chat Model, Anthropic Chat Model and Google Gemini Chat Model.
How it runs
Trigger: When a chat message arrives.
- Set debate settings (Edit Fields (Set)): sets 7 fields: question, domain, maxRounds…
- Debater (AI Agent): runs an AI agent, with model: Debater model: OpenAI
- Reviewer (AI Agent): runs an AI agent, with model: Reviewer model: Anthropic
- Parse verdict (Code): runs custom JavaScript code
- Reviewer agrees? (If): checks a condition and branches into true / false
- Judge (AI Agent): runs an AI agent, with model: Judge model: Google Gemini
- Rounds left? (If): checks a condition and branches into true / false
- Parse decision (Code): runs custom JavaScript code
- Add reviewer feedback (Code): runs custom JavaScript code
- Judge approved? (If): checks a condition and branches into true / false
…plus 4 more steps. See the full node list below.
Services it uses
Before you upgrade to n8n 3.0
Nothing in this workflow is removed in n8n 3.0, so it should import and run after the upgrade.
Nodes used (9)
- Chat Trigger
@n8n/n8n-nodes-langchain.chatTrigger - Edit Fields (Set)
n8n-nodes-base.set - No Operation, do nothing
n8n-nodes-base.noOp - AI Agent
@n8n/n8n-nodes-langchain.agent - OpenAI Chat Model
@n8n/n8n-nodes-langchain.lmChatOpenAi - Anthropic Chat Model
@n8n/n8n-nodes-langchain.lmChatAnthropic - Code
n8n-nodes-base.code - If
n8n-nodes-base.if - Google Gemini Chat Model
@n8n/n8n-nodes-langchain.lmChatGoogleGemini
From the author
Quick overview Ask a question in chat and get an answer that survived review: one model drafts it, a second challenges it, and a third rules on the result, either approving the answer or replacing it with its own corrected version. How it works 1. Receives a…