Self-Hosted LiveKit Dashboard
Stop Paying
Per Minute.
Keep the Profits.
The self-hosted LiveKit admin panel that replaces command lines with clicks. Own your data. Run local AI models. Give clients their own login. No coding required.
Why I Built This
If you’re building Voice AI in the US, life is easy. But for the rest of us in Australia, Europe, or Asia, we’re fighting physics.
Before your AI even “thinks,” you’ve lost 300ms sending audio to an American server. Add in the processing time, and the “conversational” magic breaks.
-
300ms+ latency before AI processing even begins -
Managing LiveKit servers via SSH and command lines -
Digging through logs to debug voice agent issues -
No visual interface for agent management
I needed to host locally but manage globally. So I built the solution I needed.
Latency Comparison
With on-shore models via AWS Bedrock Sydney or Google Vertex AI Australia
Why This Dashboard is Different
A stateless, ultra-lightweight dashboard that gives you GUI control without the database bloat.
Engineered for Non-US Latency
Specifically select Geo-Located Models like Gemini in Sydney or Bedrock in London to keep processing on-shore. Stateless architecture adds zero latency to the call loop.
Goodbye Command Line
Visual editor for prompts, temperature, and voice models. Hot-swap configurations while agents run. Live debugging shows real-time transcription and tool usage.
Smart Flux Integration
Deepgram Flux solves the “Interruption Problem.” Turn-complete events understand context, so agents wait politely when users pause to think.
Built for Agencies
White-label sub-users for clients. They see your logo, their agents, and their recordings. You keep admin control. Cost tracking for real-time margin analysis.
Comprehensive Feature Set
Everything you need to build, deploy, and manage production voice AI infrastructure.
Ultra-Low Latency Architecture
Speed is the only metric that matters. Built on a zero-bloat, stateless architecture that talks directly to the LiveKit API.
-
✓
Geo-Located Model Selection – Tag and select “On-Shore” models vs “Global” endpoints to shave off 250-300ms of transit latency -
✓
Latency X-Ray – Visualize P50, P90, and P99 latency metrics broken down by STT, LLM, and TTS processing times -
✓
Stateless & Lightweight – No heavy databases slowing you down. Designed for maximum speed
Visual Agent Control Center
Stop managing agents via SSH and command lines. Start, stop, edit, and duplicate agents with a click.
-
✓
No-Code Management – Visual interface for all agent operations -
✓
Deepgram Flux Integration – Native support for “Turn-Complete” events for complex data entry -
✓
Smart Handoffs – Route calls between agents with Full Context, Summarized Context, or Fresh Start options -
✓
Hot-Swapping – Change system prompts and tool definitions while the agent is running
Agency & Client Management
Built for agencies running voice AI solutions. Give clients transparency while keeping admin control.
-
✓
White-Label Sub-Users – Clients get their own login to view agents, recordings, and analytics -
✓
Granular Permissions – Restrict clients to “View Only” or specific agents/phone numbers -
✓
Cost Tracking – Input your LLM/STT token costs for real-time margin analysis and billing reports -
✓
Your Branding – Clients see your logo, not the system backend
Real-Time Observability
See everything happening in your voice AI infrastructure. Debug issues before they become problems.
-
✓
Live Room Persistence – Keep “Ghost Rooms” visible for 5+ minutes after calls end to debug events -
✓
Server Health Monitor – Localized resource tracking for Dashboard, LiveKit, Redis, and SIP services -
✓
AI Insights – Auto-generated summaries and outcome classifications for every call -
✓
Live Debugging – Click into any active room to see real-time transcription and tool usage
Infrastructure & Security
Sleep soundly knowing your infrastructure is protected. Built-in security features without the complexity.
-
✓
Hybrid Cloud Storage – Keep dashboard local, offload recordings to AWS S3 or Google Cloud Storage -
✓
Integrated Fail2Ban – GUI-based security that auto-bans IPs scanning for vulnerabilities -
✓
IP Whitelisting – Add your office IP with one click so you never lock yourself out -
✓
One-Click Updates – See which packages have updates and install them instantly
Smart Flux Integration
Deepgram Flux understands context. Old models rely on silence timers.
Interrupts. “Got it! So your email is peter@…”
Waits politely because it knows you’re not done.
See It In Action
Full control without the command line. Manage your entire voice AI infrastructure from one dashboard.
Join the High-Performance
Voice AI Community
Don’t settle for “good enough” demos. Build infrastructure that works in the real world, on real networks, for real clients.
Questions? Email info@tronic.com.au or join our Skool community for access and support.