Show HN: Knotch – a hub-and-spoke voice agent (github.com)

Knotch has introduced a novel coordination engine that manages multiple simultaneous voice channels through a content-aware AI coordinator. This innovative system stands out by efficiently determining which participants should receive specific messages in scenarios where team members cannot monitor all communications simultaneously, such as in kitchens or emergency response teams. Unlike traditional systems that rely on a human dispatcher or congested shared channels, Knotch combines the selectivity of routers with the awareness of a dispatcher to optimize communication flow, minimizing irrelevant messages and improving overall efficiency. Significantly, Knotch is designed to be domain-agnostic, allowing flexibility in implementation across various fields. With a straightforward setup involving a four-file domain structure, developers can easily adapt the engine to their specific needs without altering the core routing logic. The system utilizes a single large language model (LLM) call per utterance, ensuring deterministic output while penalizing false positives to prevent disruptions in critical environments. By facilitating real-time decision-making and self-improvement through evaluation and updates, Knotch represents a major advancement in automated communication systems, potentially transforming how teams coordinate activities across diverse domains.