Free practice first. Paid instruction when it fits.
Learn to build systems that use language models, tools and feedback to complete a task. Start with free guides and executable evaluation labs, then decide whether you need a focused module or the broader AI/ML Professional Bootcamp.
By Debasish Maji · Page reviewed October 6, 2026. Check course pages for current availability.
For developers who prefer to learn independently. Read the agent loop guide, work through examples and test a small verifier before choosing a paid program.
No enrollment required for these guides. They are not instructor-led classes or a substitute for a full production deployment.
Explore the free practice pathFor learners who already have the foundations and want the Agentic AI & Multi-Agent Systems part of the curriculum. Review its tool-use, orchestration and deployment topics.
Check its current teaching status before buying. A module marked upcoming is not a complete recording library.
Check module scope and availabilityFor learners who want the broader progression from Python and statistics through machine learning, deep learning, generative AI and agents.
This is the broader program, not an agent-only crash course. Review the published schedule, regional price and enrollment options.
View the full bootcamp curriculumPrerequisites: basic Python, JavaScript Object Notation (JSON), web application programming interfaces (APIs) and error handling. If those are new to you, explore the free visual learning resources first. Work through these four checkpoints in order; reading a page alone is not evidence of mastery.
Checkpoint 1
Explain how a model chooses a tool, reads the result and decides whether to continue. Separate a fixed workflow from a model-directed agent.
Read the free agentic AI guideCheckpoint 2
Define a documentation-search task, a narrow tool interface and a stopping rule. Model-backed examples require your own credentials and may incur provider charges.
Work through the developer tutorialCheckpoint 3
Use the dependency-free lab to detect missing updates, unintended writes and duplicate deliveries. This is a toy verifier, not a production agent.
Run the state-verification Python labCheckpoint 4
Apply quality and latency gates before comparing cost per successful task. The example numbers are synthetic, not provider benchmarks.
Run the model-selection Python labUse this checklist to evaluate any course, including ours. These are selection criteria, not a separate promised class schedule.
Tool use: Validate tool arguments and restrict actions to an explicit allowlist.
Retrieval-augmented generation (RAG): Find relevant evidence, cite it and recognize when the evidence is insufficient.
State and orchestration: Explain when a single agent is enough and when a graph or multiple agents adds useful control.
Evaluation: Check required outcomes, forbidden side effects and repeated-run failures on held-out cases.
Operational limits: Set time, step and cost budgets; handle failures and escalate instead of looping indefinitely.
Tool protocols: Understand Model Context Protocol (MCP) for tool connectivity and Agent-to-Agent (A2A) for interoperability, without confusing either with a complete agent runtime.
Build a documentation assistant over a small set of public documents. Start with one read-only search tool, return citations and say when no supporting evidence exists. Add a maximum step count and a spending limit before adding more tools.
Keep held-out questions with expected sources. Test missing evidence, a tool timeout and an instruction embedded inside a document that the agent must ignore. Record success, failures, cost and end-to-end latency. A useful portfolio includes those results and limitations, not just a polished chat screenshot.
This is a suggested practice project, not a claim about student outcomes or a job guarantee.
No. This page helps you choose among existing Thrive With AI resources: free guides, the Agentic AI & Multi-Agent Systems module, or the broader AI/ML Professional Bootcamp. Each paid option has its own published scope and availability; there is no separate 12-week enrollment offered here.
Yes. The guides and two Python labs linked here are readable without an account. The two labs run locally without a model API key. Other model-backed tutorial examples may need your own credentials and provider budget.
For a small application that uses a hosted language model, start with Python functions, dictionaries, JSON, HTTP APIs and basic error handling. You do not need to train a neural network first. The broader bootcamp includes foundations for learners who also want machine learning and deep learning.
Do not assume that. The module catalog shows current teaching status and dates. Completed modules provide published resources; upcoming or in-progress modules depend on the scheduled classes and subsequent publication of recordings.
Use the module catalog or professional-course page for the applicable regional price and purchase terms. Review the refund policy before paying. This guide does not promise an unconditional refund, a job, a salary or a certification outcome.
Check the module catalog, full course details and refund policy. For a fit or availability question, contact us before purchasing.