Give fans a direct line
to your expertise
An AI concierge trained on your content. Fans call or chat, get answers from what you've said. 24/7, in your voice, under your control. Use our hosted service or self-host the open source code.
MIT
App License
<1s
Response Time
BYOK
Your API Keys
$0
Platform Fees
Two Ways to Use EchoTube
Hosted or Self-Hosted
Use our hosted service for the easiest setup, or self-host the open source code for full control. The app is MIT licensed — you can fork, modify, and white-label it however you want.
Full Transparency
Audit every line of code. No hidden telemetry, no black boxes. You know exactly how your data flows.
No Lock-in
Switch AI providers, host anywhere, export your data anytime. You control your stack.
Community-Driven
Built by creators, for creators. Contribute features, report issues, shape the roadmap.
What's Open vs. External APIs
You bring your own API keys. Swap providers anytime — even run local LLMs with Ollama.
Technology
How It's Built
Modern stack, designed for flexibility. Swap out any component you want.
| Component | Default | Alternatives |
|---|---|---|
| AI / LLM | Claude, OpenAI | Ollama, llama.cpp, any OpenAI-compatible API |
| Voice Calls | Vapi.ai | Optional — text chat works without it |
| Voice Cloning | ElevenLabs | Optional — use stock voices or skip |
| Database | Supabase | Self-hosted Postgres |
| App Code | 100% open source (MIT) — fork it, own it forever | |
Run it 100% locally if you want. Cloud services are optional conveniences.
Features
Everything you need
A complete platform for creators who want to give fans access to their expertise.
Voice Calls
Fans can call and ask questions by voice. Your AI responds in real-time using your content.
Text Chat
Text-based chats for fans who prefer typing. Same AI personality, works without voice APIs.
Fan Memory
Your AI remembers context. It recalls past conversations so fans don't have to repeat themselves.
Bulk Content Import
Paste your channel URL and import all your videos at once. Also works with podcasts and websites.
Unanswered Questions
See what fans ask that your AI can't answer. Add new content to fill the gaps.
Safe & Controlled
Set boundaries, block topics, and maintain your brand. Full control over what your AI can discuss.
Analytics Dashboard
Track engagement trends, popular questions, content gaps, and peak activity hours.
Website Auto-Import
Enter your website URL and we'll scrape it automatically. Set periodic refresh schedules.
Content Management
Track content inventory, processing status, and embedding coverage across all your training data.
Fully Extensible
MIT licensed. Fork it, modify it, white-label it. Build your own features.
Simple Sharing
Get a unique link to share with fans. They can start calling immediately from any device.
How It Works
Get started in minutes
From clone to live AI concierge in just a few steps.
Clone & Deploy
Clone the repo, add your API keys, and deploy to Vercel or self-host with Docker.
Import Content
Paste your YouTube channel URL. Select which videos to import as training data.
Configure AI
Set your AI's personality, boundaries, and safety settings. Test in the sandbox.
Share with Fans
Share your unique link. Fans can start chatting or calling immediately.
Get Started
Deploy in Minutes
One command to run your own instance. Bring your own API keys.
# Clone the repo
git clone https://github.com/yourusername/echotube.git
# Configure environment
cp .env.example .env.local
# Start with Docker
docker compose up -d
# Running on http://localhost:3000
You provide:
- AI API key (OpenAI, Claude, or Ollama for local)
- Vapi API key (optional, for voice calls)
- Supabase project (or self-host Postgres)
You get:
- Full control over your data
- No platform fees ever
- Unlimited creators & personas
- White-label for your brand
Architecture
How It Works Under the Hood
Modern stack, designed for performance
All processing happens in under 1 second for natural conversation flow
Built with
FAQ
Frequently asked questions
What does open core mean?
The EchoTube app is 100% open source under MIT license — you can fork, modify, and white-label it. However, it relies on external APIs (OpenAI/Claude for AI, Vapi for voice) that you pay for directly. You can run local LLMs with Ollama to reduce costs.
Isn't this just a deepfake / fake version of me?
No. EchoTube is a Q&A assistant that searches your existing content to answer fan questions. It's clearly disclosed as AI. It doesn't pretend to be you—it's a tool that helps fans find answers from what you've already said.
Can I run this completely locally?
Yes! Use Ollama or llama.cpp for local AI, self-host Postgres for the database, and skip voice calls if you only need text chat. Zero external dependencies if you want.
How does the AI learn my content?
Paste your YouTube channel URL and we import all your videos automatically. We extract transcripts and use RAG (Retrieval-Augmented Generation) so your AI answers based on what you've actually said.
What if the AI says something wrong?
You have full control. Set refusal topics, add custom instructions, and review all conversations. If your AI can't answer something, it can tell the fan to reach out directly.
Can I white-label this for my own brand?
Absolutely. MIT license means you can modify, rebrand, and even sell it. No attribution required (though we'd appreciate a star on GitHub!).