💡 Inspiration

The project was inspired by the idea of connecting multiple AI and automation tools to create a truly hands-off intelligent assistant — something that could understand a request, process it through different specialized systems, and respond naturally using synthesized speech. I wanted to explore how modular AI services like Runware, Bolt, n8n, and ElevenLabs could work together to make something more dynamic than any single system alone.

🧱 How I Built It

The system works as a small orchestration pipeline:

n8n acts as the workflow engine. It receives input (text or commands) through a webhook, then routes it to other components.

Runware handles reasoning and natural-language understanding. It interprets the user’s message, breaks it into structured intents, and returns a response.

Bolt (a lightweight agent framework) helps manage state and logic between turns, letting me add simple tool-use behavior.

ElevenLabs provides high-quality voice synthesis. Once n8n gets the agent’s final text, it converts that into speech and returns both the text and an audio stream.

All of this runs inside a minimal Dockerized environment so I can deploy or restart any service independently.

🧠 What I Learned

Workflow automation: n8n makes it surprisingly easy to connect APIs and build logic visually.

API coordination: Getting Runware, Bolt, and ElevenLabs to speak a common data format taught me a lot about JSON schema design.

Latency optimization: Chaining AI services adds delay; I learned to parallelize requests and stream partial results.

Voice synthesis tuning: ElevenLabs’ voice parameters (stability, clarity, style) can completely change the perceived personality of the assistant.o

Built With

  • bolt
  • elevenlabs
  • github
  • n8n
  • runware
Share this project:

Updates