The BAIC Agentic AI course is a live online programme for building AI systems that plan steps, call tools, use memory and complete bounded tasks. You learn the engineering behind reliable agents: orchestration, retrieval, human approval, evaluation and deployment.
THE STACK, EXPLAINED
Technologies you will work with
Understand what each tool does, where it fits and how to explain your choices.
01Python, JSON & HTTP APIs
Write the agent, validate its inputs and connect it to external services.
02LLM APIs & structured outputs
Give the model tools, validate returned arguments and separate generated text from executable actions.
03LangGraph
Model multi-step workflows as stateful graphs with branching, checkpoints and human approval.
04LangChain
Connect model calls, retrieval and reusable tool interfaces.
05Embeddings & Chroma
Retrieve relevant documents and build a searchable knowledge layer for the agent.
06MCP — Model Context Protocol
Understand client/server tool discovery, tool contracts and permission boundaries.
07FastAPI, Docker & Git
Expose an agent through an API, package the application and version the work.
08Tracing, pytest & evaluation datasets
Inspect execution paths and test task completion, tool failures, latency and cost.
YOUR LEARNING SEQUENCE
From foundations to a working project.
01
Agent foundations & setup
Distinguish a chatbot, a fixed workflow and an agent. Set up Python, environment variables, Git and model API calls. Define a bounded task and its acceptance criteria.
What you build: Build a first agent loop with a clear stopping condition.
02
Tools, function calling & validation
Define tool schemas, validate JSON arguments and connect read-only APIs. Add timeouts, retries and explicit errors. Keep model suggestions separate from approved actions.
What you build: Connect two tools and handle invalid arguments.
03
Memory, retrieval & context
Separate conversation history, workflow state and durable memory. Prepare documents, create embeddings and retrieve context with source references.
What you build: Add a searchable knowledge base and test an unsupported question.
04
Planning & orchestration
Use LangGraph nodes, edges and shared state. Practise task decomposition, checkpoints, branching and bounded multi-agent handoffs.
What you build: Resume a workflow from saved state after human review.
05
Safety, MCP & human control
Inspect prompt injection risks, least-privilege tools, execution limits and approval gates. Explore MCP tool interfaces and isolate code or web execution in controlled labs.
What you build: Block an unapproved write and record why it was denied.
06
Evaluation & production delivery
Create repeatable task-level tests, inspect traces and track latency and cost. Serve with FastAPI, package with Docker and document failure recovery.
What you build: Demonstrate the support operations agent against its acceptance tests.
THE WORK YOU CAN SHOW
A tool-using support operations agent
Build an agent that reads a support request, retrieves relevant documentation, calls a sample service and prepares a proposed action. Require human approval before a write, retain an execution trace and handle a failed tool call.
What goes into your portfolio
A working Python agent and documented tool contracts
A LangGraph workflow with saved state and human approval
A test set covering incorrect inputs, tool failures and unsupported requests
A containerised API, execution traces and a project walkthrough
What technologies are taught in the Agentic AI course?
The learning stack covers Python, LLM APIs, structured outputs, LangGraph, LangChain, embeddings, Chroma, MCP, FastAPI, Docker, Git and pytest. The focus is how these components work together in a tested agent application.
Do I need coding experience?
Basic Python, HTTP APIs and JSON are the starting point. If these are new to you, discuss Python for AI Engineering with an adviser before joining this programme.
How is Agentic AI different from Generative AI?
Generative AI produces outputs such as text or structured answers. Agentic AI adds a controlled loop that selects tools, tracks state and works through steps towards a task. Our Generative AI course focuses on grounded applications; this course focuses on tool-using workflows.
Will I build a multi-agent system?
You practise bounded agent handoffs and shared task state, alongside single-agent workflows. You also learn when a simpler single-agent or fixed workflow is the better design.
What is the Agentic AI course fee?
The Agentic AI course fee is ₹14,999, including applicable taxes. Ask a course advisor for payment details and the next batch dates.
How are the sessions delivered?
Sessions are live online. Ask a course adviser about the current teaching schedule and project scope.
Your next step starts with a conversation.
Tell us your experience and what you want to build. Ask about the syllabus, batch dates and fees.