Stress-Test Plans: Gemini vs Claude Comparison n8n Workflow

Stress-Test Your Plans with Gemini vs Claude—Before You Act

Run an n8n workflow that stress-tests your decision: Gemini plays devil’s advocate with a Plan B, while Claude weighs both options and recommends A, B, or a practical mix. Get a clear second opinion as a form report or as JSON for downstream automation.

What this workflow does

• Accepts inputs from an n8n Form (logged-in users only) or from another workflow via Execute Workflow. Required fields are plan, goal, and context.
• Neutralizes prompt injection: a Code node wraps each user text section in quoted blocks and neutralises block markers inside it, so user instructions are treated as content rather than commands.
• Gemini reviews as devil’s advocate: it identifies what Plan A overlooks, proposes a Plan B, and calls out the biggest risk—all returned in a fixed JSON schema.
• Claude synthesizes the outcome: Claude compares Plan A vs Plan B and returns a fixed-schema recommendation (A, B, or a mix), including reasoning, next steps, strengths, weaknesses, and open questions.
• Failover behavior: if Gemini fails, Claude takes over both roles. If Claude fails, the report is marked INCOMPLETE and provides no recommendation.
• Delivery options: form users see the formatted report; calling workflows receive the result as JSON.

Use cases

• Review product, staffing, or roadmap decisions before launching changes
• Stress-test operational plans (SOP updates, incident response steps, rollout strategies)
• Give SaaS teams an “AI second opinion” before committing to a risky approach
• Power agent workflows with a structured decision output (A/B/mixed)

Technical details

• Integrations: Google Gemini API (Gemini Devil’s Advocate model) and Anthropic API (Claude Synthesis model)
• Core nodes: if, Code, Form, Form Trigger, Execute Workflow, No Op, Sticky Note
• Execution: ~1 minute per run (uses both APIs)