Master Future Tech with Agentic AI Course

Program Highlights

Course Overview:

We're in an era where AI doesn't just assist—it acts. Agentic AI is redefining autonomy in machines, opening doors to applications once thought impossible. This course is your pathway to becoming part of that transformation.

ExcelR's -hour Agentic AI course will help you step confidently into the future of autonomous AI systems. Whether you're aiming to become a Generative AI engineer, LLM developer, or AI agent specialist, this programme gives you the practical foundation to succeed.

You'll learn to build intelligent AI agents that can make decisions, interact with tools, and operate independently using cutting-edge frameworks such as LangChain, AutoGPT, and advanced prompt strategies. With expert-led classes, projects, and real-world case studies, we will help you master the skills needed to lead in the evolving world of Agentic AI.

Agentic AI Corporate Training

Unlock innovation with the agentic AI corporate training program. Designed for forward-thinking organisations, this hands-on course equips your teams with cutting-edge skills in autonomous intelligent systems and real-time decision-making, empowering your workforce to lead in an AI-driven world.

Why ExcelR for Agentic AI Corporate Course

Enterprise-Aligned Curriculum
Customisable Modules for Teams
Domain-Specific Agent Use Cases
Expert-Led Delivery by Practitioners
Business-Integrated Capstone Projects
Post-Training Support

Top-Notch Faculty

Exhaustive Course Curriculum

Job Readiness

Real-life Projects

Skills Covered

AI Architecture Design

Agentic RAG Implementation

LangChain Framework

AutoGPT & Open-Ended Task Execution

Chain of Thought Reasoning

Tools and Technologies

LangChain

Pinecone

ChatGPT

Ollama

Chroma

Projects

Project 1: TrialTrack – Agentic AI for Clinical Trial Matching

Domain: Pharmaceutical R&D and Healthcare
Skill Focus: Agentic RAG Implementation

Real-Life Scenario:

Pharmaceutical companies and hospitals often need to match patients with global clinical trials based on specific criteria, such as disease condition, stage, molecular marker, or location. Manually searching through ClinicalTrials.gov, CTRI, or EU-CTR databases is time-consuming and prone to errors.

TrialTrack automates this process by combining Retrieval-Augmented Generation (RAG) and agent-based task management to deliver precise, filtered clinical trial matches in real time.

How the Skill is Applied:

Agentic RAG Pipeline
- Agent Role: Clinical Query Interpreter
- Tool: LlamaIndex + LangChain RetrieverChain
- The agent first parses the user's question (e.g., "HER2-negative breast cancer trials in India, Phase 2") using an LLM-powered intent recogniser.
- A query embedding is generated and matched against vectorised clinical trial descriptions in ChromaDB.
Multi-Document Retrieval with Context-Aware Scoring
- Tool Use: MultiVectorRetriever via LangChain + ContextualCompressionRetriever
- Retrieves matching trials while compressing irrelevant segments like trial contact details and archiving dates.
Structured Summarisation Agent
- Summarises each trial's details using a chain of:
  - load_document ➝ split_text ➝ embed ➝ retrieve ➝ LLM summariser
  - Adds structured outputs, including: Trial Name, Status, Location, Inclusion Criteria, and Sponsor.
Memory & Filtering
- Skill Applied: Context Management
- Long-term memory (via Pinecone or Chroma) ensures that if the same user returns with a slightly different query (e.g., "breast cancer Phase 3 trials"), the context is reused, reducing latency and API calls.

Learning Takeaways:

Students learn how to build domain-specific RAG pipelines using large language models (LLMs) with retrieval layers that are both medically accurate and efficient.
They understand how to embed structured documents, compress noisy metadata, and render result explanations using a controlled prompt-engineering framework.
Demonstrates how RAG is not just QA—it’s a workflow augmentation mechanism.

Project 2: FinMemo – AI-Based Investment Memo Generator

Domain: Finance / Wealth Advisory
Tool Focus: AutoGen

Real-Life Scenario:

Wealth managers prepare comprehensive investment memos for clients, tailored to their financial goals, risk appetite, market trends, and asset allocation preferences. These memos require summarising news, assessing performance, and generating compliant reports.

FinMemo uses AutoGen to coordinate a team of autonomous financial agents, each with specialised reasoning and tool use, to deliver high-quality, contextual investment reports.

How AutoGen Is Applied:

Multi-Agent Role Setup
- Supervisor Agent: Breaks down the prompt into subtasks: news research, financial performance, and risk assessment.
- Research Agent: Uses APIs (Yahoo Finance, News API) to extract company metrics and current events.
- Risk Agent: Applies volatility models and sentiment analysis using Relevance AI.
- Writer Agent: Utilises persona-based prompt templates to generate the memo (e.g., conservative vs. aggressive investors).
Persistent Task Memory
- AutoGen’s memory backend stores company and user profile histories.
- This allows the system to handle evolving tasks like:
  - "Update last week's Adani memo with recent Q4 results and add bond exposure suggestion."
Dynamic Back-and-Forth Collaboration
- Agents are not linear. The Supervisor reroutes tasks:
  - If sentiment is low, it prompts the Risk Agent to revise the strategy.
  - If research is outdated, it asks the Research Agent to re-query APIs.
Report Validation and Rewriting
- AutoGen’s inner loop ensures the final memo is reviewed and validated for structure, tone, and compliance using a secondary language checker agent.

Learning Takeaways:

Students master how AutoGen enables true open-ended reasoning, unlike static chains.
They learn how to build a team of goal-driven agents, each of which is independently callable, retry-capable, and memory-aware.
Most importantly, they build a business-deployable agentic workflow—something wealth firms can use.

Project 3: CargoFlow – Exception Handler for Logistics Operations

Domain: Logistics / SCM Automation
Tool Focus: LangGraph

Real-Life Scenario:

In real-world logistics, delays, route deviations, customs bottlenecks, or damaged goods create exceptions that must be handled urgently. Current systems can only alert users but not resolve issues.

CargoFlow uses LangGraph to model a state-aware, event-driven AI agent that handles exceptions through automatic re-routing, stakeholder updates, and predictive resolution paths.

How LangGraph Is Used:

State-Based Agent Workflow
- LangGraph nodes represent shipment states:
  - In Transit, Delayed, Route Deviated, Customs Hold, Delivered, Cancelled.
- Each state is connected to downstream actions and monitored by GraphAgentRunner.
Trigger-Based Transition Logic
- E.g., if a shipment stays in InTransit for 48 hours past ETA, it triggers a move to Delayed, followed by:
  - Query alternate carriers via the logistics API
  - Notify vendor and client via Slack integration.
  - Log root cause into ERP database (Mongo or MySQL)
Edge-Specific Tools
- Transitions between LangGraph nodes activate tools:
  - In RouteDeviated, a geolocation anomaly detector calls external GPS APIs
  - In CustomsHold, an escalation agent triggers document review or invokes human escalation.
LangGraph Memory
- Agent memory persists between state transitions, allowing it to remember the actions taken, SLAs breached, and escalation paths used for the same shipment ID.

Learning Takeaways:

Students learn how LangGraph brings deterministic agent orchestration to complex, stateful systems.
They understand event-driven programming in AI contexts—how to represent real-world objects (like shipments) as dynamic state graphs.
Discover how to integrate symbolic logic with LLM-powered decision-making, a rapidly emerging hybrid AI approach.