Interactive Mentor-Led Cohorts

(02) 8902 0900contact@aumtrend.comWhatsApp / Viber
AUM Training
HomeCourses
Who We Serve
About UsContact
LMS Portal
AGENTIC AI
20 Hours (10 Days × 2 Hours)
View Full Curriculum

Agentic AI, AI Agents & Multi-Agent Systems

By the end of this course, participants will be able to design, develop, and deploy production-grade Agentic AI and multi-agent systems using LangChain, LangGraph, and Model Context Protocol (MCP) with robust enterprise safety guardrails.

Live Interactive Syllabus
20 Hours (10 Days × 2 Hours)
•
Instructor-Led Online
•
10 Hands-on Coding Labs
What you'll master in Level 4
Target Competency Outcome

By the end of this course, participants will be able to design, develop, and deploy production-grade Agentic AI and multi-agent systems using LangChain, LangGraph, and Model Context Protocol (MCP) with robust enterprise safety guardrails.

Training Objectives & Core Deliverables

Understand AI Agent anatomy, agent loops, reasoning patterns, and autonomous execution.

Implement function calling, tool schemas, and structured error-handling workflows.

Build cyclic graph workflows, state persistence, and Human-in-the-Loop reviews with LangGraph.

Architect multi-agent collaboration systems using supervisor and parallel agent patterns.

Develop custom Model Context Protocol (MCP) servers exposing enterprise tools and resources.

Design, evaluate, and deploy an Enterprise Multi-Agent AI Assistant with safety guardrails.

Day-by-Day Training Schedule & Live Labs
10 Live Sessions • 100% Hands-on
Day 1
Module 1 – Agentic AI Fundamentals
Live Session
Topics Explored:
Limitations of standalone LLM applications
What is an AI Agent? What is Agentic AI?
LLM vs Chatbot vs AI Agent vs Agentic System
Evolution: ML → DL → GenAI → LLM Apps → AI Agents → Multi-Agent Systems
Anatomy of an AI Agent: Goal, Perception, Reasoning, Planning, Memory, Tools, Action, Feedback
Agent Types: Reactive, Goal-based, Planning, Tool-using, Autonomous, Multi-agent
The Agent Loop: Observe → Reason → Plan → Act → Observe → Repeat
Hands-on Lab Exercise

Analyze limitations of standalone LLMs and construct a foundational observe-reason-plan-act agent loop.

Day 2
Module 2 – Tool Calling and AI Agent Workflows
Live Session
Topics Explored:
What are tools and why agents need tools
Function calling & tool schemas with structured input/outputs
Tool selection by an LLM & handling tool errors
Common Tools: Calculator, Weather API, Search, Database, File system, REST APIs, Python functions
Agent Reasoning: ReAct concept (Reasoning + Action)
Agent loops & planning vs execution
Deterministic vs LLM-driven workflows
Memory: Short-term memory, conversation state, long-term memory introduction
Hands-on Lab Exercise

Build a tool-calling AI agent equipped with calculator, search, and API execution tools with structured error recovery.

Day 3
Module 3 – Langchain and LangGraph
Live Session
Topics Explored:
Why LLM frameworks are needed: Models, Prompts, Tools, Chains, Agents, Retrievers
Why graph-based agent workflows? (LangGraph architecture)
State, Nodes, Edges, Conditional edges, Agent loops, Persistence
Hands-on Lab Exercise

Construct a stateful LangGraph workflow with nodes, conditional edges, loop cycles, and state persistence.

Day 4
Module 3 – Langchain and LangGraph
Live Session
Topics Explored:
Agent Workflow pattern: User Request → Planner → Researcher → Analyzer → Response
Human-in-the-Loop: Human approval, review, and revision before sensitive actions
Hands-on Lab Exercise

Build a Research Agent Workflow: Researcher → Writer → Reviewer → Human Approval → Final Response.

Day 5
Module 4 – Multiagent Systems and MCP
Live Session
Topics Explored:
Why multiple agents? Specialized agents & agent collaboration
Agent communication & shared state: Sequential, Parallel, Supervisor architecture
Supervisor Agent pattern: Supervisor → Research | Data | Writing → Reviewer → Final Response
Hands-on Lab Exercise

Implement a collaborative multi-agent architecture using the Supervisor pattern to coordinate specialized subagents.

Day 6
Module 4 – Multiagent Systems and MCP
Live Session
Topics Explored:
Model Context Protocol (MCP): Why MCP? Architecture (Host, Client, Server)
MCP Components: Tools, Resources, Prompts
MCP Workflow: AI Agent → MCP Client → MCP Server → External Tool/Data → Result
Hands-on Lab Exercise

Set up an MCP host and client environment to query structured resources and trigger remote tools.

Day 7
Module 4 – Multiagent Systems and MCP
Live Session
Topics Explored:
Enterprise Examples: Database access, File systems, CRM, HR systems, Enterprise knowledge bases
Hands-on Lab Exercise

Build a simple MCP server exposing tools (employee lookup, weather, currency) and connect to an AI agent.

Day 8
Module 5 – Enterprise Agentic AI and Capstone
Live Session
Topics Explored:
Designing Enterprise Agents: Architecture, responsibilities, tool selection, workflow, state, memory
Agent Safety & Governance: Guardrails, tool permissions, prompt injection, data privacy, human oversight
Agent Evaluation: Task completion, tool selection accuracy, response quality, groundedness, observability
Production Considerations: Reliability, latency, cost, monitoring, logging, error handling, scalability
Hands-on Lab Exercise

Implement safety guardrails, permission boundaries, and observability tracing for an enterprise agentic pipeline.

Day 9
Module 5 – Enterprise Agentic AI and Capstone
Live Session
Topics Explored:
Enterprise Multi-Agent AI Assistant Capstone architecture
Multi-agent state synchronization and orchestration
Integration of MCP tools, vector databases, and human review gates
Hands-on Lab Exercise

Capstone Project: Enterprise Multi-Agent AI Assistant (Architecture, Tool Setup & Agent State Orchestration).

Day 10
Module 5 – Enterprise Agentic AI and Capstone
Live Session
Topics Explored:
Capstone Project Finalization & End-to-End Testing
End-to-end multi-agent evaluation, resilience stress tests, and deployment packaging
Hands-on Lab Exercise

Capstone Project: Enterprise Multi-Agent AI Assistant (Final Integration, Testing & Live Demonstration).

Track Portfolio Capstone Project

Enterprise Multi-Agent AI Assistant Capstone: A collaborative multi-agent system orchestrating specialized supervisor, researcher, analyzer, and writer agents with custom MCP tools, vector memory, human approval gates, and production observability.

Detailed Module-wise Curriculum
5 Modules

Limitations of standalone LLM applications

What is an AI Agent? What is Agentic AI?

LLM vs Chatbot vs AI Agent vs Agentic System

Evolution: ML → DL → GenAI → LLM Apps → AI Agents → Multi-Agent Systems

Anatomy of an AI Agent: Goal, Perception, Reasoning, Planning, Memory, Tools, Action, Feedback

Agent Types: Reactive, Goal-based, Planning, Tool-using, Autonomous, Multi-agent

The Agent Loop: Observe → Reason → Plan → Act → Observe → Repeat

Lab Practical: Analyze limitations of standalone LLMs and construct a foundational observe-reason-plan-act agent loop.

What are tools and why agents need tools

Function calling & tool schemas with structured input/outputs

Tool selection by an LLM & handling tool errors

Common Tools: Calculator, Weather API, Search, Database, File system, REST APIs, Python functions

Agent Reasoning: ReAct concept (Reasoning + Action)

Agent loops & planning vs execution

Deterministic vs LLM-driven workflows

Memory: Short-term memory, conversation state, long-term memory introduction

Lab Practical: Build a tool-calling AI agent equipped with calculator, search, and API execution tools with structured error recovery.

Why LLM frameworks are needed: Models, Prompts, Tools, Chains, Agents, Retrievers

Why graph-based agent workflows? (LangGraph architecture)

State, Nodes, Edges, Conditional edges, Agent loops, Persistence

Agent Workflow pattern: User Request → Planner → Researcher → Analyzer → Response

Human-in-the-Loop: Human approval, review, and revision before sensitive actions

Lab Practical: Build a Research Agent Workflow: Researcher → Writer → Reviewer → Human Approval → Final Response.

Why multiple agents? Specialized agents & agent collaboration

Agent communication & shared state: Sequential, Parallel, Supervisor architecture

Supervisor Agent pattern: Supervisor → Research | Data | Writing → Reviewer → Final Response

Model Context Protocol (MCP): Why MCP? Architecture (Host, Client, Server)

MCP Components: Tools, Resources, Prompts

MCP Workflow: AI Agent → MCP Client → MCP Server → External Tool/Data → Result

Enterprise Examples: Database access, File systems, CRM, HR systems, Enterprise knowledge bases

Lab Practical: Build a simple MCP server exposing tools (employee lookup, weather, currency) and connect to an AI agent.

Designing Enterprise Agents: Architecture, responsibilities, tool selection, workflow, state, memory

Agent Safety & Governance: Guardrails, tool permissions, prompt injection, data privacy, human oversight

Agent Evaluation: Task completion, tool selection accuracy, response quality, groundedness, observability

Production Considerations: Reliability, latency, cost, monitoring, logging, error handling, scalability

Enterprise Multi-Agent AI Assistant Capstone architecture

Capstone Project implementation, testing and defense

Lab Practical: Capstone Project: Enterprise Multi-Agent AI Assistant.
AGENTIC AI COMPETENCY BADGE
Certificate of Technical Competency — Agentic AI, AI Agents & Multi-Agent Systems

Complete all live hands-on labs to earn this verifiable competency badge. This credential unlocks progression to Prompt Engineering & Reusable Template Libraries.

LIVE ONLINE COHORT
Admissions Open
Enroll in Agentic AI, AI Agents & Multi-Agent Systems
Reserve your seat for upcoming live batches, syllabus consultation & fee structure.
Selected Course Program *
100% Privacy. Official Admissions Desk.

Frequently Asked Questions

Frequently Asked Questions: Generative AI Training Plan

Find clear answers to common questions about our live interactive classes, cohort schedules, project mentoring, and career assistance.

This program is designed for engineers, developers, and tech professionals looking to master applied Machine Learning, Deep Learning, Generative AI, autonomous Agentic AI systems, and practical workplace prompt engineering.

You can choose to enroll in the complete 5-track master program (recommended for end-to-end expertise) or consult our academic team to join specific tracks matching your current skills.

Track 1: Scikit-learn, MLOps, ML Pipelines. Track 2: TensorFlow, Keras, CNN, NLP. Track 3: LangChain, Chroma/FAISS, RAG, Fine-Tuning. Track 4: LangGraph, MCP, ReAct, Multi-Agent Systems. Track 5: Prompt Engineering, Reusable Template Libraries & Risk Management.

Yes, all classes are live, interactive online sessions with senior practitioners. Session recordings, code notebooks, datasets, and reading references are shared after each class.

Still have questions? We're here to help!

Speak directly with our technical advisors to discuss cohorts, curriculum, and career assistance.

FREE 1-ON-1 ACADEMIC GUIDANCE

Ready to Start Your Tech Journey?

Connect with our academic training counsellors. We analyze your background, career goals, and recommend the best live online track for your success.

Call Helpline: (02) 8902 0900WhatsApp or Viber Support