AI Frameworks

​

Why use xpander with existing frameworks?

While frameworks like SmolAgents, LlamaIndex, and CrewAI provide powerful agent capabilities, they often lack the production infrastructure needed for real-world deployment. xpander.ai enhances these frameworks with:

• Persistent State Management: Scale to thousands of concurrent users while maintaining complex agent state across sessions
• Enhanced Agentic Tools: Access a unified toolkit of powerful connectors and integrations that work across frameworks:
  • Automatically generated AI connectors from API docs or HTTP traffic analysis
  • Pre-built integrations with popular services (LinkedIn, YC, Hacker News, etc.)
  • Framework-specific tool adapters that maintain consistent behavior across providers
  • Tool composition capabilities for building complex multi-step workflows
• Human Interfaces & Authentication: Easily connect your agents to Slack, Teams, REST APIs, voice interfaces, or hosted WebUIs like Chainlit with minimal configuration, while managing secure user authentication and access control
• Security & Compliance: Deploy on-premises within your organization’s VPC for complete data control

xpander.ai doesn’t replace your favorite agent framework — it supercharges it with production-ready capabilities.

​

Implementation Examples

Choose this approach when you want to leverage existing agent frameworks with added xpander.ai capabilities like state management, authentication, and enhanced tools.

query = "Hello, world! You are AI Agent with State managed by xpander.ai. You are now have access to more tools, authenticate users, and preserve state, return to tasks later"

smolagent = CodeAgent(
		    tools=[xpander_agent.get_tools("smolagents")],
		    model=LiteLLMModel(model_id=xpander.get_model()),
		)
smolagent.memory = xpander_agent.get_memory(agent.to_dict())
result = smolagent.run(query)
xpander_agent.send_result(result)

​

Event streaming (Slack, Teams, RestAPI, Realtime voice)

def handle_event(event):
		smolagent = CodeAgent(
		    tools=[xpander_agent.get_tools("smolagents")],
		    model=LiteLLMModel(model_id=xpander.get_model()),
		)
		smolagent.memory = xpander_agent.get_memory(agent.to_dict())
    result = smolagent.run(event.payload)
    xpander_agent.send_result(result)

xpander_agent.start_event_listener()

​

Next Steps