Google and Kaggle open enrollment for a free 5-day AI agents course

Google and Kaggle are relaunching their free five-day online course on AI agents, and registration for the new cohort opened on April 27, 2026. The intensive itself runs from June 15 to 19. It is a return of a format the companies already tested late last year: according to Google, the first AI Agents Intensive Course drew more than 1.5 million learners. Now the program is being relaunched with updated content, new speakers and a mandatory hands-on finale in the form of a capstone project, so that participants do not just watch materials but take their own idea to a working result.

The course lasts five days and is held entirely online. Google describes it as an intensive that takes learners from basic principles to production-ready agent systems, meaning solutions that can be not only assembled for a demo but also prepared for more serious use. Registration is free, so the barrier to entry has been deliberately kept minimal: sign up and set aside time to complete the program. For Google and Kaggle, it is also a way to turn AI agents from a fashionable talking point into a mass applied skill for developers, analysts, product managers and anyone already working with generative AI.

The main theme of the intensive is so-called vibe coding. In the course description, Google uses the term for an approach in which natural language becomes the primary programming interface: the user states a task in text, refines the constraints, checks the result and builds the needed logic step by step. Importantly, the course promises to cover not only quick demo scenarios but full working processes, where an agent can plan actions, call tools, connect to external APIs and operate as part of a larger system. This shift from a single prompt to a resilient process is the program's core educational goal.

According to the announcement, each day will combine conceptual walkthroughs with practical examples. That matters because the AI course market is crowded with either general lectures or scattered tutorials with no architectural picture. Here the emphasis is on linking theory and building: participants are promised a look at how to create more effective agent systems by connecting tools and APIs, then consolidating everything in a final project. The program is also stressed to be updated compared with the November launch, so participants can expect fresher use cases and examples closer to the real tasks of teams already adopting an agent approach in products and internal processes.

The relaunch has a broader context. Over recent months, AI agents have fully moved past buzzword status and become a distinct engineering discipline: companies are testing autonomous and semi-autonomous scenarios, developers are building assistants on top of LLMs, and product teams are looking for ways not just to generate text but to automate actions. Against this backdrop, Google benefits not only from promoting its own tools but also from teaching audiences practical patterns for working with agents. Kaggle, in turn, provides a familiar educational wrapper and an established platform for mass online learning, where interest in the topic can quickly be turned into real practice.

What this means in practice: the market is quickly normalizing agent development as a baseline skill rather than an experiment for enthusiasts. For participants, it is a convenient entry point: a free format, a short timeframe, clear dates and a focus on practice reduce the risk of getting stuck in abstract theory. If the updated program really delivers more applied scenarios and a strong final project, the course will be useful not only to beginners but also to those already experimenting with LLM applications who want to move from single prompts to systems with tools, APIs and clear task-execution logic.