AgentPhone Is Now on LangChain

LangChain powers over 100,000 AI applications. More than 1,300 companies use it in production, from startups to enterprises. It's been downloaded over 130 million times. And as of today, any LangChain agent can make phone calls, send messages, and manage phone numbers with a single pip install.

We're excited to announce langchain-agentphone, our official LangChain integration package. It's live on PyPI and documented on the LangChain site.

Why LangChain Matters

LangChain isn't just popular. It's become the connective tissue of the AI agent ecosystem. With 134,000 GitHub stars, it's one of the most-starred open source projects in history. A recent State of Agent Engineering survey found that 57.3% of professionals already have agents running in production. The framework supports everything from simple chains to complex multi-agent orchestration through LangGraph.

LangChain's integration ecosystem is what makes it powerful. Hundreds of tool integrations — from vector databases to search APIs to payment systems — plug into LangChain agents through a clean, standardized interface. When your agent needs a new capability, you add a tool. That's it.

Phone calls and messaging have been conspicuously absent from that list. Until now.

What You Get

The langchain-agentphone package gives your LangChain agent 12 tools covering the full phone lifecycle:

• Send messages to any phone number from your agent's own number
• Make AI-powered phone calls with full voice conversations and automatic transcripts
• Buy and manage phone numbers with specific area codes, right from your agent
• List calls, conversations, and contacts for complete communication history
• Create phone agents with custom instructions and voice configurations

Every tool follows LangChain's BaseTool pattern with full Pydantic schemas, so your LLM knows exactly what arguments to pass. No guessing. No prompt hacking.

Get Started in 60 Seconds

Install the package

pip install langchain-agentphone

Set your API key

export AGENTPHONE_API_KEY="your_api_key_here"

Give your agent a phone

from langchain_agentphone import AgentPhoneToolkit
from langchain.chat_models import init_chat_model
from langgraph.prebuilt import create_react_agent
 
model = init_chat_model(
    model="claude-sonnet-4-6",
    model_provider="anthropic",
)
 
toolkit = AgentPhoneToolkit(
    selected_tools=["send_sms", "make_call", "list_numbers"]
)
 
agent = create_react_agent(model, toolkit.get_tools())
 
response = agent.invoke({
    "messages": [
        ("user", "Send a text to +14155551234 saying 'Your order is ready for pickup!'")
    ]
})

That's it. Your LangChain agent can now text customers, make phone calls, and manage its own phone numbers. No Twilio. No webhook server. No telephony infrastructure.

Use Only What You Need

Not every agent needs all 12 tools. The toolkit's selected_tools parameter lets you expose only the capabilities your agent should have:

# A notification agent that can only send messages
toolkit = AgentPhoneToolkit(selected_tools=["send_sms"])
 
# A support agent that handles calls and reads transcripts
toolkit = AgentPhoneToolkit(
    selected_tools=["make_call", "get_transcript", "list_calls"]
)
 
# Full access — all 12 tools
toolkit = AgentPhoneToolkit()
tools = toolkit.get_tools()

You can also import individual tools directly if you want even more control:

from langchain_agentphone import AgentPhoneSendSMS, AgentPhoneMakeCall
 
send_sms = AgentPhoneSendSMS()
make_call = AgentPhoneMakeCall()
 
agent = create_react_agent(model, [send_sms, make_call])

What You Can Build

LangChain agents with phone access open up use cases that were previously out of reach:

Order Notification Agent

Connected to your e-commerce backend, it texts customers when their order ships, calls them if delivery fails, and handles "where's my package?" inquiries over SMS. No human in the loop.

Appointment Reminder Agent

Pulls tomorrow's schedule from your calendar API, sends personalized reminder texts, and calls no-shows to reschedule. Healthcare, salons, consulting — any business that loses revenue to missed appointments.

Lead Qualification Agent

When a form submission comes in, the agent calls the lead within seconds. It asks qualifying questions, answers product inquiries using RAG over your docs, and books a meeting with your sales team. Speed to lead, automated.

Part of a Growing Ecosystem

This LangChain integration joins our existing Google ADK integration and MCP server that works with Claude Code, Cursor, Windsurf, and any MCP-compatible client. Whether you're building with LangChain, ADK, or connecting directly through MCP, AgentPhone gives your agent a phone.

LangChain's upcoming Interrupt conference (May 13-14, 2026) is bringing together the agent engineering community to discuss what's next. With 57% of teams already running agents in production, the question has shifted from "should we build agents?" to "what capabilities do they need?" Telephony is high on that list.

Get Started

The integration is live on PyPI today. Here's how to dive in:

• Install the package
• Read the LangChain docs
• Explore the full API reference
• Get your API key