Google and Kaggle Are Bringing Back Their AI Agents Course — Here’s What You Need to Know

Google and Kaggle are bringing back their free AI Agents Intensive Course for a second round, running June 15-19, 2026. Registration opened today, and if last year’s numbers are anything to go by, spots will fill up fast. The first iteration hit over 1.5 million learners, which is impressive even by Google’s standards.

This time around, the focus is squarely on what they’re calling “vibe coding” — essentially building AI agents using natural language as your primary interface. If you’ve been playing with tools like Claude or GPT-4 and wondering how to move from chat-based experiments to actual production systems, this course is aimed directly at you.

The format is five days of online sessions, mixing conceptual deep dives with hands-on coding. They’ve updated the content since last November, added new speakers, and there’s a capstone project at the end where you actually build something. No cost to register, which is refreshing in a landscape where every AI startup seems to charge for basic tutorials.

What I find interesting is the emphasis on “10x agents” — their phrase, not mine. The idea is that by connecting tools and APIs through natural language workflows, you can create agents that punch way above their weight. It’s a bold claim, but Google has the infrastructure to back it up with things like Vertex AI and the Gemini ecosystem.

Of course, there’s a catch. Vibe coding is still early. You’re essentially describing what you want in English (or whatever language) and hoping the model translates that into working code. It works surprisingly well for simple tasks, but I’ve seen it fall apart on anything that requires precise logic or edge-case handling. The course will likely teach you where it shines and where you need to drop down to traditional programming.

If you’re already comfortable with Python and have messed around with LangChain or similar frameworks, this might feel a bit basic. But if you’re new to agent development and want a structured path from zero to deployed system, it’s hard to beat free and backed by the people who built the underlying tech.

Registration is on the course website. No idea if they cap enrollment, but given the first run’s popularity, I wouldn’t dawdle.