Listen to this Post
LangChain has introduced the Open Agent Platform (OAP), a no-code solution for building, deploying, and orchestrating LangGraph agents through a web-based interface. This platform is designed for both non-technical users and advanced teams working with AI agents.
Key Features:
🔹 Visual Agent Management – Create, configure, and interact with LangGraph agents directly from a browser.
🔹 Three Agent Types:
1. Standard Agents – For single-purpose tasks.
- Tools Agents – Access external services via MCP (Multi-Component Protocol).
3. Supervisor Agents – Coordinate multiple agents.
🔹 RAG Integration via LangConnect – Agents can connect to retrieval-augmented generation servers for external data reasoning.
🔹 Authentication & Access Control – Built-in via Supabase, supporting Google login or custom providers.
Infrastructure Requirements (Self-Hosting):
- LangSmith account
- Supabase project
- LangGraph agent deployment
- MCP-compatible server (e.g., Arcade)
- LangConnect (optional but supported)
- LLM API key (OpenAI, Anthropic, Google)
Setup Flow:
1. Develop agents with LangGraph.
2. Deploy on the LangGraph Platform.
3. Configure deployments in OAP.
- Connect tools via MCP and knowledge via LangConnect.
5. Launch and manage through the web UI.
🔗 GitHub Repo: https://github.com/langchain-ai/open-agent-platform
You Should Know:
1. Setting Up LangGraph Locally
To experiment with LangGraph agents, install the required packages:
pip install langgraph langchain-openai
2. Running a Basic Agent
Here’s a simple Python script to create a LangGraph agent:
from langgraph.graph import Graph
from langchain_core.messages import HumanMessage
from langchain_openai import ChatOpenAI
Initialize LLM
llm = ChatOpenAI(model="gpt-4-turbo")
Define agent workflow
workflow = Graph()
workflow.add_node("agent", lambda x: llm.invoke(x))
workflow.set_entry_point("agent")
workflow.set_finish_point("agent")
Run the agent
response = workflow.invoke(HumanMessage(content="Explain how LangGraph works."))
print(response)
3. Deploying with Supabase
To enable authentication, set up Supabase:
npm install @supabase/supabase-js
Configure environment variables:
export SUPABASE_URL="your-project-url" export SUPABASE_KEY="your-anon-key"
4. Integrating LangConnect for RAG
Use LangChain’s RAG pipeline:
from langchain_community.vectorstores import FAISS
from langchain_openai import OpenAIEmbeddings
Load documents and create a vector store
documents = ["LangGraph is a framework for multi-agent workflows."]
vectorstore = FAISS.from_texts(documents, OpenAIEmbeddings())
Retrieve relevant info
retriever = vectorstore.as_retriever()
docs = retriever.invoke("What is LangGraph?")
print(docs)
5. Managing Agents via CLI
List running agents (if using a local server):
curl -X GET http://localhost:8000/agents
What Undercode Say:
The Open Agent Platform significantly lowers the barrier to AI agent development, making it accessible to non-coders while offering advanced customization for developers.
Expected Linux/Windows Commands for AI Agent Management:
- Check running AI agent processes (Linux):
ps aux | grep "langgraph"
- Kill a misbehaving agent (Linux/Windows):
kill -9 $(pgrep -f "agent_name") Linux taskkill /IM "agent_process.exe" /F Windows
- Monitor API calls (Linux):
sudo tcpdump -i any port 8000 -A
- Deploy via Docker (for self-hosting OAP):
docker-compose up -d
Prediction:
As no-code AI agent platforms evolve, we’ll see more businesses adopting autonomous agent workflows for customer support, data analysis, and automated decision-making.
Expected Output:
A functional LangGraph agent deployed via OAP, accessible via a web interface, with integrated RAG and multi-agent coordination.
References:
Reported By: Shivanivirdi Langchain – Hackers Feeds
Extra Hub: Undercode MoN
Basic Verification: Pass ✅
