n8n Workflow: Batch Structured AI Webhook with JSON Schema

Batch Structured AI Webhook for n8n (JSON Schema + Ordered Results)

Give your AI coding agent a single authenticated webhook that accepts a batch of prompts and JSON Schemas—then get back ordered, structured outputs from the first 50 calls processed in parallel.

What this workflow does

• Receives batched requests: A Webhook accepts a POST request with an X-API-Key header and a calls array. Each call includes systemPrompt, userMessage, a JSON-encoded jsonSchema string, and an aiMode of small or big. Use one mode per request.
• Checks authentication: The workflow looks up the token in the users Data Table. Unknown tokens return HTTP 401.
• Logs usage: An Insert log stores the user’s email and the full submitted calls payload (prompts, schemas, and modes).
• Limits and parallelizes: After Prepare Calls and Split Calls, it Limit(s) to 50 items—processing the first 50 in parallel and silently dropping extras. If you need more, split jobs and verify result counts.
• Routes models + enforces schema: The Model Selector routes small to GPT 6 Luna and big to Gemini 3.8 Flash. A Structured Output Parser applies each call’s JSON schema.
• Returns ordered results: Respond to Webhook returns {query, output} entries—one per processed call—preserving the input order.

Use cases

• Running AI-assisted code generation or refactors where every response must match a strict JSON Schema.
• SaaS operators batch-validating structured outputs (e.g., extracted fields, tool-ready payloads) with traceable logging.
• Automation engineers orchestrating high-throughput LLM batches with a single webhook endpoint.

Technical details

• Nodes / logic: if, set, no op, limit, webhook, split out
• Authentication: X-API-Key validated against users Data Table
• Batch behavior: processes first 50 calls; extras are dropped
• Model routing: small → GPT 6 Luna, big → Gemini 3.8 Flash
• Structured parsing: schema-driven output per call