Microsoft Azure User Group - Chennai

Modern AI demos often look impressive, but many are simply LLM prompts stitched together.

Real Agentic AI systems are different.

They require structured memory, orchestration, and tool-driven agents working together like a coordinated system.

In this hands-on workshop, we will explore how to design production-ready Agentic AI architectures instead of simple prompt chains.

Participants will learn how to build a system where:

• An Orchestrator Agent acts as the reasoning brain
• Specialized task agents perform actions
• Tools execute real operations like data queries and calculations
• Memory systems store context, decisions, and knowledge

Using LangChain, we will build a simple but powerful architecture with multiple agents working together.

The system will include:

• An Orchestrator Agent that understands the user request and coordinates tasks.
• A Data Agent that retrieves and processes structured data using tools
• A Decision Agent that analyzes results and produces recommendations
• An Ambiguity Detection Agent that identifies unclear questions and asks for clarification before the system makes decisions

This approach demonstrates an important principle:

LLMs should act as the reasoning brain, not the entire system.
We will also explore how memory works inside agentic systems
including:

• Working memory for current tasks
• Episodic memory for past reasoning steps
• Semantic memory for persistent knowledge

By the end of the session, participants will understand how to design scalable Agentic AI systems with proper architecture rather than ad-hoc prompts.

***

# Who Should Attend

• Software Engineers
• AI Engineers
• Architects building intelligent systems
• Developers curious about multi-agent AI architectures

***

# What You’ll Learn

• How agent orchestration works
• Why memory is critical in AI systems
• When agents should use tools instead of LLM reasoning
• How to design multi-agent architectures using LangChain
• How to detect ambiguity before taking action

***

⚡ Key takeaway:
The future of AI systems is not just better models.
It is better architectures built around reasoning, tools, and memory.