SIPHON

Siphon is an open-source Python framework that abstracts the complexity of building AI calling agents, letting developers ship voice AI in hours instead of weeks. THE DEVELOPER PROBLEM: Building AI calling agents requires deep expertise in: • SIP signaling and telephony infrastructure • Real-time WebRTC audio streaming • Orchestrating LLM, STT, and TTS services with sub-500ms latency • Managing interruptions, state, and async event handling • Implementing call persistence and monitoring Most dev teams spend 2-4 weeks on infrastructure before writing conversation logic. THE SOLUTION: Siphon provides a batteries-included framework. Here's a complete AI receptionist: from siphon.agent import Agent from siphon.plugins import openai, cartesia, deepgram agent = Agent( agent_name="receptionist", llm=openai.LLM(), tts=cartesia.TTS(), stt=deepgram.STT(), system_instructions="You are a helpful receptionist..." ) if __name__ == "__main__": agent.start() # Handles real PSTN calls That's production-ready. Deploy it and it answers phone calls. ARCHITECTURE: • Built on LiveKit (WebRTC transport layer) • Native SIP integration for PSTN connectivity • Streaming audio pipeline for <500ms voice-to-voice latency • Plugin system for AI providers (OpenAI, Anthropic, Groq, etc.) • Horizontal auto-scaling with zero configuration • Automatic transcript/recording/metadata persistence TECHNICAL HIGHLIGHTS: • Vendor-agnostic: Swap LLM/STT/TTS providers via config • Self-hosted: Apache 2.0 license, you own the infrastructure • Production-ready: Built-in monitoring, error handling, state management • Developer-friendly: Type hints, clear APIs, comprehensive docs PERFECT FOR BUILDING: • Customer support automation • Appointment scheduling systems • Sales qualification pipelines • Healthcare intake workflows • Any telephony + AI use case WHY WE BUILT IT: After months spent wrangling SIP infrastructure for a voice AI project, we realized every developer building calling agents hits the same walls. Siphon removes these barriers. REPO & DOCS: GitHub: https://github.com/blackdwarftech/siphon Docs: https://siphon.blackdwarf.in/docs Install: pip install siphon-ai