Build AI Agents That Connect Everywhere

A unified API to enable your AI agent to talk to users via email, Slack, SMS, WhatsApp, custom clients and more.

Multi-Channel Access

Enable your AI agents to talk to users across multiple communication platforms with zero extra code.

OAuth Management

We take care of authenticating and managing OAuth tokens for your users.

Analytics

Track usage and interactions between your agent and users.

Easy human in the loop

Reach out to your users in their preferred communication channel to get consent for agent actions.

Join the Waitlist

Be among the first to access our platform

Talk to users via any channel

Easily integrate with multiple communication channels. Your AI agent can receive and respond to messages from any platform through a single unified callback.

from agentreach import agentreach

def callback(
    session_id,
    message,
    past_conversations,
    user_id,
    communication_channel
):
    print(f"Received message from {user_id} via {communication_channel}: {message}")
    response_from_llm = call_llm(message)
    user_response = agentreach.send_message_to_user_and_get_response(
        session_id,
        response_from_llm
    )
    # TODO: ... continue conversation

    return agentreach.end_session(session_id)

agentreach.register_callback(<AGENT_ID>, callback)

Manage OAuth tokens for your users

Let us handle the complexity of OAuth. Simply request tokens for the services you need, and we will manage the authentication flow with the user's consent and token refresh cycles.

from agentreach import agentreach

def callback(
    session_id,
    message,
    past_conversations,
    user_id,
    communication_channel
):
    # Get an OAuth access token for this user that has calendar access
    access_token = agentreach.get_oauth_token(
        session_id,
        "google",
        ["https://www.googleapis.com/auth/calendar"]
    )
    # TODO: Use access_token to access Google Calendar API

agentreach.register_callback(<AGENT_ID>, callback)

Get user consent effortlessly

Implement human-in-the-loop workflows with a single function call. Get user approval before taking important actions while maintaining the conversation flow.

from agentreach import agentreach

def callback(
    session_id,
    message,
    past_conversations,
    user_id,
    communication_channel
):
    # ...
    user_consent = agentreach.get_consent_from_user(
        session_id,
        "I am scheduling an event on your calendar with the following details: ..."
    )
    if not user_consent:
        return "User did not consent to take action"

    # ...