01 · Challenge
Conversational voice platforms may send slightly different payload shapes, which can break downstream automation if the integration assumes a single format.
Conversational AI · Blocking webhooks
An ElevenLabs webhook tool and n8n workflow that receives caller messages in blocking mode, normalizes payloads, and routes lead capture, lookup, and general AI-response branches.

Case study brief
01 · Challenge
Conversational voice platforms may send slightly different payload shapes, which can break downstream automation if the integration assumes a single format.
02 · Response
I built an ElevenLabs-to-n8n workflow that accepts blocking webhook calls, normalizes payload variants, and returns concise results for lead capture, lookup, or general assistance.
System anatomy
The architecture is expressed as responsibilities and boundaries so the model, workflow logic, interfaces, and human controls remain inspectable.
ElevenLabs sends caller messages to a blocking webhook tool
The workflow normalizes payload variants into a stable JSON contract
Action branches handle lead capture, lookups, or a general AI response
Implementation
Defined the ElevenLabs webhook tool and expected response contract.
Mapped alternative caller-message fields into one stable JSON structure.
Separated lead, lookup, and AI-response branches with explicit routing.
Returned compact voice-ready output and retained human escalation as a safe fallback.
Evidence & validation
Two reusable agent/workflow configuration exports.
Voice-agent screenshot and documented integration structure.
Dedicated project post and evidence assets.
Reflection
Integration quality improves when connector-specific variability is absorbed at the boundary rather than spread across every branch.
Next iteration
Outcomes
Importable n8n workflow structure
Reliable payload normalization across caller-message variants
Concise voice-ready responses with defined escalation behavior