Agent communication protocols in Agentic AI

Introduction

Agent communication protocols help different AI agents talk and work together smoothly, like friends sharing ideas to solve a problem.

When multiple AI agents need to share information to complete a task.

When coordinating robots to work together in a factory.

When chatbots need to pass messages to each other to answer complex questions.

When AI systems in a smart home communicate to manage devices efficiently.

When agents negotiate or make decisions together in a game or simulation.

Syntax

Agentic AI

protocol = {
    'message_type': ['inform', 'request', 'confirm', 'query'],
    'sender': 'agent_name',
    'receiver': 'agent_name',
    'content': 'information or command',
    'timestamp': 'time_sent'
}

Each message follows a clear structure to avoid confusion.

Common message types include 'inform' to share facts and 'request' to ask for actions.

Examples

AgentA tells AgentB that the door is open.

Agentic AI

message = {
    'message_type': 'inform',
    'sender': 'AgentA',
    'receiver': 'AgentB',
    'content': 'The door is open',
    'timestamp': '2024-06-01T10:00:00Z'
}

AgentB asks AgentA to close the door.

Agentic AI

message = {
    'message_type': 'request',
    'sender': 'AgentB',
    'receiver': 'AgentA',
    'content': 'Please close the door',
    'timestamp': '2024-06-01T10:01:00Z'
}

AgentA confirms to AgentB that the door is closed.

Agentic AI

message = {
    'message_type': 'confirm',
    'sender': 'AgentA',
    'receiver': 'AgentB',
    'content': 'Door is now closed',
    'timestamp': '2024-06-01T10:02:00Z'
}

Sample Model

This program shows two agents sending and receiving messages using a simple communication protocol. Each message has a type, sender, receiver, content, and timestamp.

Agentic AI

import time

class Agent:
    def __init__(self, name):
        self.name = name

    def send_message(self, receiver, message_type, content):
        message = {
            'message_type': message_type,
            'sender': self.name,
            'receiver': receiver.name,
            'content': content,
            'timestamp': time.strftime('%Y-%m-%dT%H:%M:%SZ', time.gmtime())
        }
        print(f"{self.name} sends to {receiver.name}: {message}")
        receiver.receive_message(message)

    def receive_message(self, message):
        print(f"{self.name} received message: {message}")

# Create two agents
agent_a = Agent('AgentA')
agent_b = Agent('AgentB')

# AgentA informs AgentB
agent_a.send_message(agent_b, 'inform', 'The door is open')

# AgentB requests AgentA
agent_b.send_message(agent_a, 'request', 'Please close the door')

# AgentA confirms to AgentB
agent_a.send_message(agent_b, 'confirm', 'Door is now closed')

Important Notes

Timestamps help agents know when messages were sent.

Clear message types avoid misunderstandings between agents.

Protocols can be extended with more message types as needed.

Summary

Agent communication protocols let AI agents share messages clearly.

Messages have a type, sender, receiver, content, and time.

Using protocols helps agents work together like a team.