* Matching an agent from whatever a speech model said. Client-safe and pure, because both halves need it: `askAgent` resolves on the server, while `activateAgent` is handled inline in the voice hook and resolves in the browser. One implementation keeps the two from disagreeing ab
// Source: https://github.com/keinsaasforever/better-chatbot | License: MIT | Copyright 2025 better-chatbot
// Provided via justclaw.it.com; keep this notice with the code.
/**
* Matching an agent from whatever a speech model said.
*
* Client-safe and pure, because both halves need it: `askAgent` resolves on
* the server, while `activateAgent` is handled inline in the voice hook and
* resolves in the browser. One implementation keeps the two from disagreeing
* about which agent "the sales one" means.
*/
export type AgentChoice = { id: string; name: string };
/**
* Find the agent a reference points at, or `null`.
*
* A speech model does not say UUIDs. Even with the roster in its instructions
* it paraphrases — "the sales agent", "Sales", "unser Vertriebsagent" — so an
* id-onl
... (truncated -- full source via MCP)
See the full source, get the GitHub permalink, and search for more like it.
Get a free API key