Google and Kaggle did run a free five-day AI Agents: Intensive Vibe Coding Course from June 15 through June 19, 2026. But the claim that the course “pulled in 353,000 learners worldwide” does not hold up against the public record. Google has not published a post-course enrollment figure for the June program, Kaggle’s public listing does not show that number, and the only clearly matching 353,000 figure found in current search results is an estimate of Kaggle’s YouTube subscriber count.
That correction changes the story. Google’s latest developer-education effort is real, completed, and still useful for people who want hands-on exposure to agent development. It is not, however, documented as a 353,000-person training milestone. The public evidence supports a narrower conclusion: Google reran a free agent-building course after its 2025 predecessor attracted more than 1.5 million learners, but it has not yet disclosed whether the 2026 edition approached anything close to that scale.
The original reporting from TechBuzz appears to have conflated an unsupported enrollment count with the course’s wider visibility. For Windows developers and IT teams evaluating Google’s growing agent stack, the more important takeaway is that this was a developer on-ramp to Google’s tooling—not evidence that hundreds of thousands of people completed a production-agent curriculum.
Google announced the course on April 27, 2026, calling it the return of its AI Agents Intensive with Kaggle. Registration was free, the live intensive ran June 15–19, and Google framed the updated version around vibe coding: using natural language as a primary programming interface while building agent systems that call tools and APIs.
The five-day event has since concluded. Google’s developer-profile system still lists its associated course badge, and Google Codelabs identifies at least one secure-agent development exercise as part of the program. Kaggle’s competitions directory also continues to list the event as a featured hackathon associated with the June dates.
What is absent is just as important. Neither Google’s April launch announcement nor the publicly indexed Kaggle listing provides an enrollment total for the 2026 run. Google has published no recap comparable to its December 2025 post about the first AI Agents Intensive. That earlier recap reported more than 1.5 million learners, more than 160,000 people in the Discord community, 2 million views of course whitepapers, 3.3 million notebook views, and more than 11,000 capstone submissions.
Those were unusually specific metrics. No corresponding numbers have been released publicly for the June 2026 course.
Kaggle’s currently indexed competitions page lists the event with “0 Teams.” That should not be read as zero learners; a team counter on a hackathon page is not an enrollment or completion counter. It does demonstrate, however, that the page is not publicly reporting 353,000 participants. A live-course enrollment total, a badge total, notebook views, Discord membership, capstone submissions, and hackathon teams all measure different things. Treating any one of them as “developers trained” needs evidence that has not been supplied here.
The number has another problem: third-party channel-tracking pages indexed in late June placed Kaggle’s YouTube subscriber count at roughly 353,000 to 355,000. That does not prove where the course claim originated, but the match is close enough that the reported learner total looks likely to be a category error rather than a verified course metric.
Then came the first dedicated AI Agents Intensive in November 2025. Google and Kaggle said it welcomed more than 1.5 million learners. That is the documented high-water mark that the April 2026 announcement cited when it promoted the rebooted course.
Against that backdrop, an enrollment figure of 353,000 for the 2026 edition would be plausible in the abstract—but it would represent a steep decline from the 1.5 million claimed for the prior agents course. That would be newsworthy in its own right, and Google has not said it happened.
More importantly, “enrolled learners” is not the same thing as trained developers, active attendees, completers, or production-ready engineers. The 2025 recap made this distinction indirectly by publishing several separate measures: course participation, Discord activity, content views, and capstone projects. The 11,000-plus project submissions were the closest public measure of learners who converted course material into a concrete deliverable, and even that is not a certification of real-world deployment capability.
The submitted report’s assertion that 353,000 “dwarfs typical online course completion rates” is therefore unsupported twice over. It compares enrollment to completion without establishing either the program’s actual enrollment total or its completion rate.
That direction is visible in Google’s Agent Development Kit, or ADK. Google launched ADK in April 2025 as an open-source framework for developing multi-agent applications. Its current documentation describes support for Python, TypeScript, Go, and Java; local development and debugging; tool integration; orchestration; testing; and deployment to environments including Cloud Run and Google Kubernetes Engine.
Google also positions ADK as model-flexible and deployment-flexible. That is useful for Windows-centric organizations that do not want every agent project locked permanently to one model provider. But the practical center of gravity remains Google’s product line: Gemini models, Vertex AI services, Google Cloud Agent Runtime, and Google’s managed enterprise-agent platform.
The course’s “vibe coding” label should not obscure that commercial direction. Natural-language-driven coding can lower the barrier to producing a prototype, especially for a PowerShell, .NET, Python, or IT automation practitioner with a clear operational problem in mind. It does not remove the work required to make that prototype safe.
Google’s own materials emphasize evaluation, test paths, and deployment controls alongside development. That is a welcome contrast to the more careless interpretation of vibe coding—prompt until something appears to work, then ship it. An agent that can read internal knowledge, query a ticket system, create cloud resources, modify files, or invoke administrative tools needs permissions boundaries, audit trails, adversarial testing, secrets management, and a way to stop or override it.
For Windows administrators, the immediate use case is likely to be narrow agents rather than autonomous digital employees: a tool that triages support tickets, drafts but does not send remediation plans, analyzes Windows event logs, prepares patch-impact summaries, or helps build scripts under human review. The moment an agent can execute a command, change an identity permission, or act on an external API, it belongs inside the organization’s normal change-control and security processes.
The December 2025 recap of the earlier agents program explicitly said its coursework had been converted into a Kaggle Learn Guide. The April 2026 launch post promised updated content, new speakers, live sessions, and a capstone project. It did not say whether every lesson, notebook, and lab from the June run would remain publicly accessible afterward.
That gap matters more than a flashy registration count. A five-day event gives developers a useful survey of concepts, but durable training value comes from whether the labs remain available, whether their dependencies are maintained, and whether the examples keep pace with a rapidly changing agent stack.
Google may yet publish a formal recap with a verified participation number. Until it does, the defensible record is simple: the June 2026 Google-Kaggle AI Agents course ran as scheduled, it was free, and it was designed to bring more developers into Google’s agent-development workflow. The claim that it trained 353,000 people should be treated as unsubstantiated, not repeated as a milestone.
The original reporting from TechBuzz appears to have conflated an unsupported enrollment count with the course’s wider visibility. For Windows developers and IT teams evaluating Google’s growing agent stack, the more important takeaway is that this was a developer on-ramp to Google’s tooling—not evidence that hundreds of thousands of people completed a production-agent curriculum.
The June 2026 course happened, but the 353,000 figure is unverified
Google announced the course on April 27, 2026, calling it the return of its AI Agents Intensive with Kaggle. Registration was free, the live intensive ran June 15–19, and Google framed the updated version around vibe coding: using natural language as a primary programming interface while building agent systems that call tools and APIs.The five-day event has since concluded. Google’s developer-profile system still lists its associated course badge, and Google Codelabs identifies at least one secure-agent development exercise as part of the program. Kaggle’s competitions directory also continues to list the event as a featured hackathon associated with the June dates.
What is absent is just as important. Neither Google’s April launch announcement nor the publicly indexed Kaggle listing provides an enrollment total for the 2026 run. Google has published no recap comparable to its December 2025 post about the first AI Agents Intensive. That earlier recap reported more than 1.5 million learners, more than 160,000 people in the Discord community, 2 million views of course whitepapers, 3.3 million notebook views, and more than 11,000 capstone submissions.
Those were unusually specific metrics. No corresponding numbers have been released publicly for the June 2026 course.
Kaggle’s currently indexed competitions page lists the event with “0 Teams.” That should not be read as zero learners; a team counter on a hackathon page is not an enrollment or completion counter. It does demonstrate, however, that the page is not publicly reporting 353,000 participants. A live-course enrollment total, a badge total, notebook views, Discord membership, capstone submissions, and hackathon teams all measure different things. Treating any one of them as “developers trained” needs evidence that has not been supplied here.
The number has another problem: third-party channel-tracking pages indexed in late June placed Kaggle’s YouTube subscriber count at roughly 353,000 to 355,000. That does not prove where the course claim originated, but the match is close enough that the reported learner total looks likely to be a category error rather than a verified course metric.
Google’s own benchmark makes the claim harder to read at face value
Google has a credible history of using Kaggle intensives to reach enormous audiences. Its first five-day Generative AI Intensive in late 2024 drew more than 140,000 registrants, according to Google. The second GenAI Intensive in March and April 2025 drew more than 280,000 learners and set a Guinness World Record for the largest virtual AI conference, Google later said.Then came the first dedicated AI Agents Intensive in November 2025. Google and Kaggle said it welcomed more than 1.5 million learners. That is the documented high-water mark that the April 2026 announcement cited when it promoted the rebooted course.
Against that backdrop, an enrollment figure of 353,000 for the 2026 edition would be plausible in the abstract—but it would represent a steep decline from the 1.5 million claimed for the prior agents course. That would be newsworthy in its own right, and Google has not said it happened.
More importantly, “enrolled learners” is not the same thing as trained developers, active attendees, completers, or production-ready engineers. The 2025 recap made this distinction indirectly by publishing several separate measures: course participation, Discord activity, content views, and capstone projects. The 11,000-plus project submissions were the closest public measure of learners who converted course material into a concrete deliverable, and even that is not a certification of real-world deployment capability.
The submitted report’s assertion that 353,000 “dwarfs typical online course completion rates” is therefore unsupported twice over. It compares enrollment to completion without establishing either the program’s actual enrollment total or its completion rate.
The course is a route into Google’s agent platform
Google’s educational push is still strategically meaningful, even if the headline number is not. The course introduced developers to a workflow that aligns closely with Google’s commercial AI-agent offerings: build an agent around a model, connect it to tools and external APIs, test whether it behaves as intended, then deploy it with controls appropriate to production use.That direction is visible in Google’s Agent Development Kit, or ADK. Google launched ADK in April 2025 as an open-source framework for developing multi-agent applications. Its current documentation describes support for Python, TypeScript, Go, and Java; local development and debugging; tool integration; orchestration; testing; and deployment to environments including Cloud Run and Google Kubernetes Engine.
Google also positions ADK as model-flexible and deployment-flexible. That is useful for Windows-centric organizations that do not want every agent project locked permanently to one model provider. But the practical center of gravity remains Google’s product line: Gemini models, Vertex AI services, Google Cloud Agent Runtime, and Google’s managed enterprise-agent platform.
The course’s “vibe coding” label should not obscure that commercial direction. Natural-language-driven coding can lower the barrier to producing a prototype, especially for a PowerShell, .NET, Python, or IT automation practitioner with a clear operational problem in mind. It does not remove the work required to make that prototype safe.
Google’s own materials emphasize evaluation, test paths, and deployment controls alongside development. That is a welcome contrast to the more careless interpretation of vibe coding—prompt until something appears to work, then ship it. An agent that can read internal knowledge, query a ticket system, create cloud resources, modify files, or invoke administrative tools needs permissions boundaries, audit trails, adversarial testing, secrets management, and a way to stop or override it.
For Windows administrators, the immediate use case is likely to be narrow agents rather than autonomous digital employees: a tool that triages support tickets, drafts but does not send remediation plans, analyzes Windows event logs, prepares patch-impact summaries, or helps build scripts under human review. The moment an agent can execute a command, change an identity permission, or act on an external API, it belongs inside the organization’s normal change-control and security processes.
What Google still has not disclosed
Google has not announced a price because the intensive itself was free. It has also not published a cohort-completion rate, badge count, capstone total, geographic breakdown, demographic breakdown, or a deadline for releasing the course material as a self-paced Kaggle Learn guide.The December 2025 recap of the earlier agents program explicitly said its coursework had been converted into a Kaggle Learn Guide. The April 2026 launch post promised updated content, new speakers, live sessions, and a capstone project. It did not say whether every lesson, notebook, and lab from the June run would remain publicly accessible afterward.
That gap matters more than a flashy registration count. A five-day event gives developers a useful survey of concepts, but durable training value comes from whether the labs remain available, whether their dependencies are maintained, and whether the examples keep pace with a rapidly changing agent stack.
Google may yet publish a formal recap with a verified participation number. Until it does, the defensible record is simple: the June 2026 Google-Kaggle AI Agents course ran as scheduled, it was free, and it was designed to bring more developers into Google’s agent-development workflow. The claim that it trained 353,000 people should be treated as unsubstantiated, not repeated as a milestone.
References
- Primary source: The Tech Buzz
Published: 2026-08-03T15:50:08.548698
Google Trains 353K Developers on AI Agents in Free Course | The Tech Buzz
Kaggle's AI Agents Intensive drew massive enrollment as Google pushes developer educationwww.techbuzz.ai - Related coverage: blog.google
Google and Kaggle launch AI Agents Intensive course
Earlier this year, we launched the second iteration of our GenAI Intensive course, which attracted over 280,000 learners. Now, we're taking things to the next level with…blog.google - Related coverage: blog.google
Inside Kaggle’s AI Agents Intensive Course with Google
Kaggle’s AI Agents Intensive with Google brought learners together in a no-cost course to build and deploy the next frontier of AI.blog.google - Related coverage: linkedin.com
Learn about AI agents in 5 days with an interactive, no-cost course from Kaggle 🧠 Sign up for the AI Agents Intensive course from November 10 - 14 to explore the foundations and practical… | Google for Developers | 19 comments
Learn about AI agents in 5 days with an interactive, no-cost course from Kaggle 🧠 Sign up for the AI Agents Intensive course from November 10 - 14 to explore the foundations and practical applications of AI agents, giving you the skills to build your own projects efficiently. →...www.linkedin.com
- Related coverage: linkedin.com
Join the 5-day AI Agents Intensive course with Google and Kaggle. | Anant Nawalgaria
AI Agents are the next frontier. After our "5 Days of GenAI" courses broke the world record and drew over half a million learners, we are back with the next iteration—100% focused on AI Agents. Move beyond prompts. Learn to build everything from simple AI agents to sophisticated multi-agent... - Related coverage: kdnuggets.com
Kaggle + Google’s Free 5-Day Agentic AI Course - KDnuggets
Google and Kaggle's 5-Day AI agents course is now freely available to everyone.www.kdnuggets.com
- Related coverage: techscurrent.com
Google and Kaggle AI Agents Course: What Starts Today
Google and Kaggle’s free AI Agents Intensive starts June 15, 2026. Here’s what the vibe-coding course covers, who should join, and how to prepare.techscurrent.com - Related coverage: storage.googleapis.com
- Related coverage: destination-ai.tdsynnex.fr
- Related coverage: luminhkhuong.dev
AI Agents & Vibe Coding: Overview | Engineering Knowledge Base
Introduction to AI Agents, the core agentic formula, the evolution of LLMs, and why autonomous agents are transforming software development in the vibe coding era.luminhkhuong.dev - Related coverage: navigotechsolutions.com
Google AI Vibe Coding Course for Indian Marketers: Full Guide 2026 - navigotechsolutions.com
Learn how the Google AI Vibe Coding course helps Indian marketers build automation tools without coding. Practical steps, tools, and examples for small businesses.navigotechsolutions.com - Related coverage: groups.google.com
- Related coverage: kingy.ai
Best Free AI Courses in 2026: OpenAI Academy, Google/Kaggle Agents, and the AI Skill Stack
A practical 2026 guide to the best free AI courses, including OpenAI Academy, Google/Kaggle AI Agents, Hugging Face, Microsoft Learn, IBM, Anthropic, and DeepLearning.AI, plus role-based learning paths and projects.kingy.ai - Related coverage: docs.cloud.google.com
Agent Development Kit | Gemini Enterprise Agent Platform | Google Cloud Documentation
Learn how to use ADK in Gemini Enterprise Agent Platform.docs.cloud.google.com
- Related coverage: github.com
GitHub - google/adk-docs: An open-source, code-first toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control. · GitHub
An open-source, code-first toolkit for building, evaluating, and deploying sophisticated AI agents with flexibility and control. - google/adk-docs
github.com
- Related coverage: developers.google.com
GEAR | Google Developer Program | Google for Developers
developers.google.com
- Related coverage: developers.googleblog.com
Agent Development Kit: Making it easy to build multi-agent applications - Google Developers Blog
Discover Google ADK, an open-source framework for streamlined development offering precise control, rich tools, and easy deployment.developers.googleblog.com - Related coverage: codelabs.developers.google.com
Agents CLI in Agent Platform: From Development to Production | Google Codelabs
codelabs.developers.google.com
- Related coverage: developers.googleblog.com
Agent Development Kit: Making it easy to build multi-agent applications - Google Developers Blog
Discover Google ADK, an open-source framework for streamlined development offering precise control, rich tools, and easy deployment.developers.googleblog.com - Related coverage: docs.cloud.google.com
Quickstart: Develop and deploy agents on Agent Runtime with Agent Development Kit | Gemini Enterprise Agent Platform | Google Cloud Documentation
Learn how to create, deploy, and test an AI agent on Agent Runtime using the Agent Development Kit (ADK). - Related coverage: cloud.google.com
Master Generative AI Evaluation: From Single Prompts to Complex Agents | Google Cloud Blog
Master the critical step of GenAI evaluation. Learn to move your LLM, RAG, and complex agent applications from prototype to production using a rigorous, metrics-driven approach with hands-on labs for Vertex AI Evaluation and the Agent Development Kit (ADK).cloud.google.com
- Related coverage: github.com
Update llms.txt to align with the llms.txt standard and act as a sitemap for models · Issue #726 · google/adk-docs · GitHub
As per llmstxt.org the llms.txt file should ideally provide an index to the key resources in folder, site or repo. It should provide a list of links, where each link has a summary of the content that can be found by following that link. ...
github.com
- Related coverage: aiagentsagency.ca
- Related coverage: services.google.com
- Related coverage: techradar.com
Google Cloud is making its AI Agent Builder much smarter and faster to deploy | TechRadar
Vertex AI Agent Builder is getting quicker to usewww.techradar.com - Related coverage: developers.google.com
5-Day AI Agents: Intensive Vibe Coding Course With Google | Google Developer Program | Google for Developers
I've earned the 5-Day AI Agents: Intensive Vibe Coding Course With Google badge!
developers.google.com
- Related coverage: codelabs.developers.google.com
Vibecode and Secure an AI Agent Lifecycle with Antigravity and TDD | Google Codelabs
Vibe-code a secure shopping assistant agent using Agent Development Kit (ADK) and Gemini models. Secure boundaries with Antigravity IDE, Semgrep in pre-commit hooks, and a custom STRIDE threat modeling skill.
codelabs.developers.google.com
- Related coverage: skool.com
5-Day AI Agents: Intensive Vibe Coding Course With Google · Clief Notes
For anyone interested there's a 5 day course coming up June 15 - 19, 2026. https://www.kaggle.com/competitions/5-day-ai-agents-intensive-vibecoding-course-with-www.skool.com
