✨ New Feature

Your AI Companion
for Live Streams

An AI-powered avatar that watches your gameplay, reacts in real-time, and talks to your audience — all running inside OBS.

1

Pick a Companion

Each companion has a unique personality that shapes how they react to your gameplay. Choose one that fits your stream vibe.

🔷
Byte
Tactical & witty
👾
Fizz
Maximum hype
🍡
Mochi
Cozy & chill
🛡️
Pixel
Retro RPG hero
👻
Error
Nervous intern
🎀
Celest
Sweet helper
You can also write a custom prompt to create your own personality from scratch.
2

Connect a Vision AI

Your avatar needs "eyes" to see what's on screen. Pick a vision provider and add your API key.

Preset Free
Built-in vision — no key needed. Great for testing.
OpenAI GPT-4o BYOK
Excellent image understanding. Needs an OpenAI API key.
Google Gemini BYOK
Fast and capable. Free tier available from Google AI Studio.
Kimi k2p5 BYOK
Alternative vision model from Moonshot AI.
BYOK = Bring Your Own Key. Your keys are stored encrypted and never shown after saving. Enter __clear__ to remove a key.
3

Choose a Voice

Your companion can speak out loud on stream. Pick a text-to-speech provider.

Browser TTS Free
Uses your browser's built-in speech synthesis. Works everywhere, no key required.
ElevenLabs BYOK
Ultra-realistic AI voices. The best quality available — highly recommended.
OpenAI TTS BYOK
High-quality voices from OpenAI. Uses your existing OpenAI key.
Pro tip: ElevenLabs offers the most natural-sounding voices. You can customize the voice by setting a Voice ID in the Studio.
4

Add to OBS & Go Live

Your avatar appears on stream as an OBS Browser Source. Here's what it looks like:

That was a sick play!
🎀

Browser Source Setup

Copy the overlay URL from Avatar Studio, add it as a Browser Source in OBS, and position it on your scene.

400 × 300 px
Desktop Client: Download and run the StreamTracker desktop client — it captures your screen periodically and sends it to the vision AI. The companion then reacts with a comment and voice line that appears in the OBS overlay.

Ready to set up your avatar?

Head to the Avatar Studio to configure everything in one place.

🎭 Open Avatar Studio ← Back to Dashboard

Frequently Asked Questions

Do I need to pay for anything?
The avatar feature itself is free. The Preset vision provider and Browser TTS work without any API keys. For higher-quality vision and voice, you can bring your own API keys (BYOK) from providers like OpenAI, Google, or ElevenLabs — each has their own pricing and free tiers.
Are my API keys safe?
Yes. Your keys are stored securely and are never returned by the API after saving. You'll only see a masked version (e.g., sk-...xY). You can remove a key at any time by entering __clear__ in the key field.
How does screen capture work?
The desktop client runs on your computer and takes periodic screenshots (configurable from 1–10 minutes). Each screenshot is sent to your chosen vision AI, which generates a comment. The comment appears in the OBS overlay with a voice line.
Can I create my own personality?
Absolutely. Pick any companion as a base, then write a custom prompt in the Studio to override or extend their personality. You can make your avatar say anything in any style.
What is the Memory feature?
When enabled, your avatar remembers its recent comments and uses them as context for new responses. This makes conversations feel more natural — it won't repeat itself and can reference things it said earlier. You can set the memory depth from 1 to 50 messages.
Does the overlay work with Streamlabs?
Yes — any streaming software that supports Browser Sources (OBS, Streamlabs, XSplit, etc.) will work. Just paste the overlay URL and set the dimensions to 400×300.