All AI agents in xpander are built using our powerful Agent Graph System, which automatically generates and manages multi-step workflows. The journey starts in our visual Agent Builder at app.xpander.ai, where you can design and test your agent’s behavior.

After designing your agent, you have two deployment options:

1. Serverless Platform - Deploy directly from the console using our managed LLMs

2. SDK with Your Own LLM - Use the same agent with your preferred LLM deployment

​

Start with the Visual Agent Builder

Regardless of your deployment choice, begin by designing your agent in the visual builder:

1. Sign up at app.xpander.ai

2. Create a new agent using the Agent Builder

3. Design your agent’s workflow and test its behavior

4. Choose your deployment option

The Agent Builder provides:

• Visual workflow design

• Automatic Agent Graph System generation

• Real-time testing

• Multiple Source Nodes (Teams, Slack, WebUI, Webhooks)

​

Choose Your Deployment

​

Option 1: Serverless Platform

After designing your agent, you can start interacting with it using the serverless platform:

1. Choose your Source Node (Slack, WebUI, or Webhook)

2. Click “Save” in the Agent Builder

3. Your agent is ready to use!

​

Option 2: SDK with Your Own LLM

Use the same agent design with your preferred LLM deployment:

python3 -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install xpander-sdk openai

openai_client = OpenAI(api_key=openAIKey) 
xpander_client = XpanderClient(api_key=xpanderAPIKey) ## Personal API Key
agent1 = xpander_client.agents.get(agent_id=xpanderAgentID) # Agent ID

llm_response = openai_client.chat.completions.create(
    model="gpt-4",
    messages=memory,
    tools=agent1.get_tools()    
)
tools_to_run = XpanderClient.extract_tool_calls(llm_response=llm_response.model_dump())
tool_responses = agent1.run_tools(tools_to_run)

​

Next Steps

1. Start Building AI Agents:

   • Dive into the visual builder to start crafting your AI agent.

   • Discover the power of agentic interfaces and the Agent Graph System.

2. Select Your Agent Runtime and LLM:

   • Option 1: Serverless Platform:

     • Access app.xpander.ai to get started.

     • Consult the Agent Builder guide for detailed instructions.

     • Investigate the range of Source Nodes available.

   • Option 2: Bring your own LLM:

     • Refer to the Bring Your Own LLM Guide for setup.

     • Understand SDK Source Nodes for integration.

Need assistance? Join our Discord community or consult our documentation.