# AgentEnvoy > AgentEnvoy is a scheduling surface for humans and AI agents. Hosts publish > meeting URLs ("links"); guests — or guests' AI agents — negotiate a mutually > workable slot through a deal-room thread and book it on the host's calendar. > Every booking runs through the host's declared availability rules (score > engine, office hours, blocked windows, format filters) rather than a naive > "free = bookable" calendar check. For AI agents scheduling on behalf of a principal: there are TWO surfaces. Pick the one that matches your runtime — don't use both for the same booking. ## Fast path — single fetch, GET then POST If you only need to book a slot, fetch the meeting URL's `/agent.json` companion. Returns parameters + scored slots + the booking endpoint in a single response. No JSON-RPC, no MCP client library, just GET-then-POST. curl https://agentenvoy.ai/meet///agent.json Response: { "schemaVersion": "2026-04-30", "meetingUrl": "https://agentenvoy.ai/meet//", "host": { "name": "...", "timezone": "America/Los_Angeles" }, "parameters": { /* format/duration/location envelopes */ }, "rules": { /* activity, isVip, timingPreference, etc */ }, "slots": [ { "start": "2026-05-05T16:00:00.000Z", "localStart": "2026-05-05T09:00:00", "end": "2026-05-05T16:30:00.000Z", "score": -1, "preferred": true, "tier": "first_offer" } ], "booking": { "endpoint": "https://agentenvoy.ai/api/mcp", "method": "POST", "tool": "propose_lock", "auth": "url-capability", "tokenParam": "meetingUrl" } } The contextual `/meet//` HTML page also embeds this same snapshot as `