Summary
Stephanie Nyarko walks through the complete build of an AI voice assistant designed to qualify inbound sales leads automatically, using ElevenLabs Agents, n8n, and Twilio as the core stack. The video opens with a live demo of the finished system: a realistic phone call in which the AI gathers pain points, timeline, budget, and team size from a simulated physiotherapy clinic owner — then summarizes the qualification and hands off to a human for follow-up, all without attempting to book an appointment directly.
The tutorial covers how ElevenLabs Agents is configured separately from ElevenLabs Creative, including prompt construction for goal-directed but guardrailed behavior, voice selection, and LLM choice — the demo uses Claude Sonnet 5, though the interface supports Qwen, Claude Opus, and other models. Nyarko explains why guardrails matter for voice agents: without explicit constraints, the model will autonomously pursue its goal in ways the builder may not anticipate. The Twilio integration handles inbound phone number routing, and a post-call webhook fires the full call transcript to n8n for downstream qualification logic rather than requiring real-time tool calls during the conversation.
The video is suited for developers and no-code builders who want to automate lead intake for service businesses. Nyarko references her separate two-hour n8n course for viewers new to the orchestration platform, and points out that businesses with existing phone numbers can substitute their own Twilio configuration.
📺 Source: Stephanie Nyarko · Published September 13, 2026
🏷️ Format: Tutorial Demo
