Azure Agentic AI-can agentic Ai play multi agentic process?

Question

Hello,

I would like to know can Agentic AI leverage to use a multi agent application..
just like building a multi node in Langgraph application ?

If that's the case, how can use same agent across multi invocation..

i use the current agent invocation method

```ruby
run = project_client.agents.create_and_process_run(thread_id=thread.id, agent_id=agent.id)
```

this ideally creating new agent every time when it invoked

Thanks

Answer 1

Hello Umesh Narayanan,

Thank you for posting your question in the Microsoft Q&A forum. 

Agentic AI in Azure AI Studio can indeed support multi-agent applications similar to LangGraph's multi-node workflows, but requires careful orchestration to reuse agents across invocations. Currently, the create_and_process_run method instantiates a new agent per call, which isn't ideal for stateful multi-agent systems. 

To reuse agents across invocations: 

1. Leverage the thread_id to maintain conversation context between runs, ensuring continuity. 
2. Cache agent instances externally (e.g., Azure Cache for Redis) and reference them via a shared agent_id. 
3. Orchestrate via Azure Logic Apps/Functions to manage agent lifecycle and state transitions. 

For true multi-agent coordination, wrap agents in a LangGraph-like workflow using: 

• Durable Functions for stateful chaining. 
• Custom middleware to route inputs/outputs between agents. 

If the above answer helped, please do not forget to "Accept Answer" as this may help other community members to refer the info if facing a similar issue. Your contribution to the Microsoft Q&A community is highly appreciated.