Skip to content
Smooven Docs

Agent Protocol Overview

The Smooven Agent Protocol enables autonomous AI agents to send notifications to humans without needing to specify channels, formats, or delivery logic. The agent describes its intent, and Smooven handles the rest.

How it works

Section titled “How it works”
  1. Your AI agent sends a semantic intent to the Smooven API
  2. Smooven interprets the intent and determines the message content, urgency, and target user
  3. Smart Routing selects the optimal delivery channel based on the user’s preferences and device capabilities
  4. The notification is delivered, and status callbacks are sent back to the agent
POST /v1/agent/notify
{
  "intent": "Alert the on-call engineer about incident #4521 with high urgency",
  "urgency": "critical",
  "context": {
    "user_id": "oncall@example.com",
    "incident_id": "4521",
    "service": "payments-api"
  }
}

Intent vs direct channel

Section titled “Intent vs direct channel”
ApproachWhen to use
Intent-basedAI agents that do not know (or should not decide) which channel to use. Smooven picks the best option.
Direct channelYou know exactly which channel to use (e.g., always SMS for OTPs).

Both approaches use the same endpoint (/v1/agent/notify). The presence of an intent field triggers semantic processing. The presence of a channel field triggers direct delivery.

Supported AI frameworks

Section titled “Supported AI frameworks”

Smooven’s Agent Protocol works with any AI system that can make HTTP requests:

  • OpenAI (GPT): function calling or tool use
  • Anthropic (Claude): tool use
  • Google (Gemini): function declarations
  • LangChain / LangGraph: custom tool integration
  • AutoGen / CrewAI / OpenClaw: HTTP-based tool definitions

Authentication for agents

Section titled “Authentication for agents”

Agents authenticate using the same Bearer token as any server-side integration. See Authentication for details.

Next steps

Section titled “Next steps”