Everything on voice and chat AI agents — real-time speech processing, telephony and phone-system automation, conversational AI, voice UX design, and production deployment.
From the blog
Pipecat is the most flexible open voice framework: 20+ STT, 30+ TTS, 20+ LLMs, WebRTC and telephony. Here's a fully self-hosted Pipecat agent on FastAPI — no Pipecat Cloud needed.
Hume EVI 3 beats GPT-4o on empathy and naturalness in blind tests, at sub-300ms latency. When an emotion-aware voice agent is worth it, and when it is not.
Meta's January and April 2026 pricing changes, the new AI Provider policy, and what they mean for an SMB AI receptionist on WhatsApp.
How India's blockchain-based DLT registration works, why category-level DND preferences matter for AI calls, and the per-list scrubbing economics that decide whether you stay legal.
Daily.co Bots deprecated July 2025 and was succeeded by Pipecat Cloud (GA January 2026). The replacement keeps Twilio Voice native integration, Krisp VIVA noise reduction, and Daily WebRTC transport, with deeper customization and lower cost.
Matter 1.5.1 made WebRTC the official spec for smart-home cameras and two-way audio. Aqara's first Matter cameras shipped H1 2026. Here is the AI voice-control build pattern.
End-of-utterance detection is the metric that controls whether the agent feels fast and rude or slow and considerate. Here is how we measure EOU precision and recall, why semantic turn-end models beat plain VAD silence, and how we tune by vertical.
Live auctions in 2026 run an AI auctioneer assistant that calls bid increments, watches simulcast bidders, and adapts pace in real time. Here is the WebRTC + Pion + ringman bridge stack.
Children speak with shorter utterances, higher pitch, and less consistent grammar. We unpack COPPA 2026, the CHATBOT Act, age-band TTS, and the design boundary CallSphere enforces between kid and adult callers.
