In my experience, having built the infra for this, Twilio's MediaStreams and PSTN had higher latency than a simple SIP that interfaces with WebRTC, which you could connect to your backend. Curious how this changes with the OpenAI realtime, since my approach was stringing gpt-4o with Elevenlabs for the backend.
When PG wrote this essay (last year?) about climate tech, it was a deep dive into what the problems were, what the potential solutions and roadblocks could be, and who they're looking to fund in that space. This essay is "Solve cancer," "stablecoin finance," and "ml in robotics." Like, thanks for the insights, lol.