# Agentic Studying
In November 2025, greater than 1.5 million individuals signed up for a free course on constructing AI brokers. That is not a typo, and it isn’t a MOOC’s self-importance metric for registrations that by no means convert: the identical cohort filed greater than 11,000 capstone submissions. Google and Kaggle’s 5-Day AI Brokers Intensive turned, by attendance, one of many largest technical programs ever run.
The excellent news for anybody who missed the dwell week: all of it’s now free and self-paced, and there is much more to it than the hype steered.
# What It Truly Is
The course is Google and Kaggle’s follow-up to their 2024 GenAI Intensive, which itself drew 140,000-plus builders and a rerun that set a Guinness file for the most important digital AI convention. The 2025 version narrowed the main focus from generative AI broadly to brokers particularly, which is the best name given how a lot confusion nonetheless surrounds what an agent even is.
The dwell run occurred over 5 days in November 2025, however the entire thing has since been remodeled right into a self-paced Kaggle Study Information, so now you can work via it by yourself schedule. A refreshed run, themed round vibe coding, additionally went out in June 2026 for many who favor the cohort expertise. Both manner, the fabric is the draw.
# The 5 Days
Every day pairs a technical whitepaper with two hands-on codelabs constructed on Gemini and Google’s Agent Improvement Package, so that you’re studying the idea after which instantly constructing it.
Day 1 – Introduction to brokers: Agent architectures and the elemental query of when a process wants an agent in any respect versus a less complicated workflow. You construct your first agent and your first multi-agent system.
Day 2 – Instruments and interoperability with MCP: How brokers use instruments, easy methods to write customized ones, and the way the Mannequin Context Protocol (MCP) lets brokers discuss to exterior techniques. This present day additionally introduces human-in-the-loop approval for long-running operations, which is the distinction between a helpful agent and a harmful one.
Day 3 – Context engineering: Classes and reminiscence: Constructing stateful brokers that bear in mind, and giving them long-term reminiscence that persists throughout classes. That is the place most selfmade brokers quietly crumble, so it is value slowing down right here.
Day 4 – Agent high quality: Logging, tracing, metrics, and evaluating each an agent’s responses and its device use. In the event you take sooner or later severely, take this one. Measuring whether or not an agent works is the ability virtually no one teaches and everybody delivery to manufacturing desperately wants.
Day 5 – Prototype to manufacturing: The agent-to-agent protocol and deploying to a managed runtime like Vertex AI Agent Engine. The unglamorous final mile between a pocket book that impresses your workforce and a system actual customers can hit.
# Who Ought to Take It, and Methods to Get the Most from It
In the event you can write Python and you’ve got known as an LLM API, you are prepared. The course assumes programming literacy however not agent experience, which is strictly the band most practitioners sit in proper now.
One piece of recommendation from watching how these intensives are inclined to go: do not skim the whitepapers to hurry to the code. The analysis and context-engineering materials on days three and 4 is the place the sturdy understanding lives, and it is the half you may’t choose up later by copying a working repo. The codelabs educate you to construct an agent. The papers educate you why yours retains breaking.
The entire thing is free, it is out there now on Kaggle, and the capstone provides you one thing actual to level at if you’re performed. One million and a half individuals discovered 5 days for it. Yours are in all probability value the identical.
Nahla Davies is a software program developer and tech author. Earlier than devoting her work full time to technical writing, she managed—amongst different intriguing issues—to function a lead programmer at an Inc. 5,000 experiential branding group whose purchasers embody Samsung, Time Warner, Netflix, and Sony.
