#python #agentic_ai #agents #ai #ai_agents #realtime #stt #tts #video_agents #video_ai #vision_ai #voice_ai

Vision Agents is an open-source Python framework by Stream to build real-time AI agents that watch video, listen to audio, and respond instantly with low latency under 30ms. It integrates YOLO, Roboflow, OpenAI, Gemini, and 25+ tools for apps like golf coaching, security cameras detecting theft, or phone assistants. Install easily with `uv add vision-agents`, use free Stream credits, and deploy on any video network. You benefit by quickly creating smart video AI for gaming, safety, or coaching without vendor lock-in, saving time and costs on custom builds.

https://github.com/GetStream/Vision-Agents