Integrations

Buddy AI plays well with the wider agent ecosystem. The buddy.integrations package provides optional, dependency-light adapters so Buddy components interoperate with LangChain and LangGraph — use Buddy's unified models, agents, and tools where you already are, and bring LangChain tools into Buddy.

Version

Documents buddy-ai 2.2.0. Adapters are verified against langchain-core >= 0.3 and langgraph >= 0.2.

Why integrate instead of replace?

Buddy and LangGraph solve problems at different layers:

Layer	Strength
Buddy	Agent platform — 30+ models, tools, memory, RAG, teams, PULSE, Competency Engine
LangChain	Huge tool/connector ecosystem, chains, retrievers
LangGraph	Graph orchestration — cycles, checkpoints, human-in-the-loop

Rather than reimplement an orchestration engine, Buddy lets you use the best of each: Buddy agents as the capable unit of work, LangGraph as the orchestration engine, and LangChain for its connectors.

Installation

# LangChain interop only
pip install "buddy-ai[langchain]"

# LangGraph orchestration (also pulls in langchain-core)
pip install "buddy-ai[langgraph]"

The adapters import their third-party dependency lazily. Importing buddy.integrations.langchain or buddy.integrations.langgraph works even when the package is not installed; a clear, actionable error is raised only when you actually call an adapter that needs it.

Detecting availability

import buddy

print(buddy.LANGCHAIN_AVAILABLE)   # True if langchain-core is installed
print(buddy.LANGGRAPH_AVAILABLE)   # True if langgraph is installed
print(buddy.get_available_features())

What's included

LangChain (details)

Adapter	Purpose
BuddyChatModel	Use any Buddy model as a LangChain BaseChatModel
BuddyAgentTool	Expose a Buddy Agent/Team as a LangChain tool
to_langchain_tool / from_langchain_tool	Convert tools in either direction
to_buddy_messages / from_buddy_message	Convert message objects

LangGraph (details)

Adapter	Purpose
BuddyNode	Wrap a Buddy Agent/Team as a LangGraph node
add_buddy_node	Register a Buddy agent on an existing StateGraph
build_sequential_graph	Build & compile a linear multi-agent pipeline
build_default_state	Default state schema (input, output, messages)
make_competency_edge	Route to the most competent member via the Competency Engine

Runnable example

A complete, runnable walkthrough lives at examples/13_langchain_langgraph.py.