Google and Kaggle Just Reopened Their Free 5-Day AI Agents Course. Here’s What “Vibe Coding” Means and Why You Should Sign Up

Last November, Google and Kaggle ran a 5-day AI Agents intensive and over 1.5 million people showed up. That’s not a typo. One and a half million learners. It set a Guinness World Record for the largest virtual AI conference in a single week.

If you missed it, here’s the good news: it’s back.

Google and Kaggle just opened registration for the 2026 edition of the 5-Day AI Agents Intensive: Vibe Coding Course, running June 15 to 19, 2026. It’s completely free. There’s a real-world capstone project. And it’s built around a workflow that’s quietly changing how people build AI: vibe coding.

Let’s break it down.

What “Vibe Coding” Actually Means

You’ve probably seen the term floating around X and YouTube. Most explanations make it sound mysterious. It isn’t.

Vibe coding is when natural language becomes your main programming interface. You describe what you want in plain English, and the AI handles the syntax, scaffolding, and integration work. You stay in the role of architect and editor, not the person fighting curly braces at 1 AM.

The shift sounds small. It isn’t. It changes who can build software, how fast they ship, and how complex the systems get before they break.

This course teaches you to use vibe coding to build what Google calls “10x agents”: AI systems that don’t just answer questions but plan, call tools, talk to other agents, remember context, and ship real results.

What’s New in the 2026 Edition

If you took the November intensive, this is not a rerun. Google has updated the curriculum, brought in new speakers, and added a real capstone.

Here’s what’s actually different:

Vibe coding is the lens. The previous course focused on foundational agent concepts. This year leans into how natural-language workflows accelerate agent building.
An optional capstone project. Instead of just running through tutorials, you’ll design, build, and deploy your own agent system.
Certificates and badges. Successfully completing the capstone earns you an official Kaggle certificate and badge.
Prizes for the best work. Top winners walk away with exclusive Kaggle swag and social media recognition from Google.
Updated speakers. New guests sharing what’s actually working in production right now.

The capstone alone is worth showing up for. Watching tutorials is one thing. Designing and shipping your own agent is the part that changes your skill ceiling.

What You’ll Actually Be Building

Across five days, you’ll go from “I’ve heard about agents” to “I can build one.”

The curriculum covers:

Core agent components. Models, tools, orchestration, memory, and evaluation. The full stack.
External tool and API integration. Connecting your agent to real services so it can actually do things.
Agent-to-agent communication. Multiple agents working together on complex, multi-step tasks.
Long-term memory. Engineering memory so your agent remembers context across sessions.
Production-ready deployment. Going from notebook prototype to something you’d let real users touch.

Each day combines deep-dive sessions, hands-on codelabs, live Discord discussions, and YouTube livestreams with Google’s own AI engineers and researchers.

Who This Course Is Actually For

Be honest with yourself before signing up. It saves you a week of frustration.

You’ll get a lot out of this if you:

Can already write basic Python and have used the Gemini, OpenAI, or Claude APIs at least once
Want to move past chatbots and build agents that actually do work
Are a developer, data scientist, or technical PM exploring agent use cases
Like learning by building, not just watching

You’ll struggle if you:

Have never written a line of code
Are looking for a no-code, ChatGPT-prompt-tips kind of session
Hate live deadlines and codelab time pressure

If you fall in the second bucket, the previous course’s content is now a free Kaggle Learn Guide, and you can work through it at your own pace before deciding if the live intensive is right for you.

What You Need Before You Register

A few logistics so nothing surprises you on Day 1:

A free Kaggle account. The platform requires phone number verification before you can run the codelabs, so do it now and not at 6 AM on June 15.
Access to a Google AI Studio account. This is where you’ll get your API keys for Gemini.
Working Python skills. You don’t need to be advanced. But if for loop and function are mysteries to you, do a Python crash course first.
A Discord account. Most live discussion happens there. Last year, over 160,000 learners joined the conversation.
Roughly 2 to 4 hours per day, blocked on your calendar. Treat it like a meeting you can’t reschedule.

How to Register in Under Two Minutes

Here’s the exact path:

Go to the official course page on Kaggle.
Sign in or create a free Kaggle account.
Verify your phone number in your Kaggle account settings (this is mandatory for codelabs).
Click “Join” on the competition page to register.
Set up your free Google AI Studio account so you have API access ready.
Block June 15 to 19 on your calendar.
Watch your inbox for the Discord invite when it goes out.

That’s it. You’re in.

How to Actually Finish It (And Get Something Out of It)

Last year over 11,000 capstone projects were submitted. That’s a lot. But it also means hundreds of thousands of registrants didn’t ship anything. Don’t be that group.

Here’s what works, based on how the previous intensive ran:

Pick a project idea before the course starts. Think small. A research assistant agent. A meeting-prep agent. A travel planner. The capstone goes way smoother when you bring a problem in, instead of inventing one mid-week.
Read the whitepapers in the morning. Code in the evening. This pacing worked for thousands of learners last time. Don’t try to do both at once.
Lurk on Discord. Then post. The fastest unlock in any community is asking one specific question and answering one specific question. Free mentorship without paying for it.
Don’t aim for perfect. Aim for shipped. A working agent that does one thing well will teach you more than a half-built one with five features.
Save your code somewhere durable. GitHub. A personal Notion. Anywhere. Future-you will thank you when you’re updating your CV in October.

Why This One Matters Right Now

AI agents are no longer a research demo. They’re showing up in customer support, internal ops, content workflows, security automation, and product features at almost every company taking AI seriously.

The people who can actually design, build, and ship agents are about to be a small, valuable group. Free courses from the team behind Gemini and the largest data science platform in the world don’t come around twice a year.

Registration is open right now. The course runs once. Show up.

If this post helped you decide, send it to the friend or coworker who’s been saying “I really should learn this stuff” for six months. Five days from now they could be building something instead of bookmarking another article.

Source: Google’s official announcement on The Keyword and the Kaggle competition page.