AI Customer Support Workflow: OpenRouter + ElevenLabs in n8n

Turn customer messages into instant, AI-powered support—with text and voice—in n8n

This AI Customer Support Workflow uses OpenRouter for conversational replies, ElevenLabs for speech-to-text and text-to-speech, and n8n Data Tables to log every interaction—plus optional SMTP email handoff to a human when escalation is needed.

What this workflow does

• Listens for incoming requests via the n8n Chat Trigger (text) or a webhook endpoint for voice requests.
• For voice: sends uploaded audio to ElevenLabs Speech-to-Text, then normalizes text/voice inputs into a shared format with a session ID.
• Generates an AI reply using an OpenRouter chat model, using the customer message plus your company info, FAQs, and business rules, with short-term conversation memory.
• Detects escalation by checking for an escalation marker in the AI output; if present, it replaces the response with a predefined handoff message and flags the conversation for human support.
• Logs the conversation (customer message, AI reply, session ID, timestamp, and handoff flag) to an n8n Data Table.
• Optionally escalates to a human: when enabled, sends an email via SMTP including conversation details, then prepares the final customer response.
• Responds to the customer with text, or for voice requests returns MP3 audio produced by ElevenLabs Text-to-Speech.

Use cases

• Automate SaaS support for common FAQs and policy questions (text chat).
• Handle phone-style inquiries by transcribing voice, replying conversationally, and returning an MP3 answer.
• Escalate to support staff for complex issues—while keeping a searchable conversation log in n8n Data Tables.

Technical details

• Nodes/logic: if, set, code, webhook, Data Table, Email Send (SMTP).
• Integrations: OpenRouter (chat model responses), ElevenLabs (Speech-to-Text + Text-to-Speech).
• Setup essentials: add an OpenRouter API credential and choose a chat model; configure an ElevenLabs API connection.