Kaggle AI Agents Intensive is the November 2025 sequel to the Gen AI Intensive on this site. Live cohort November 10 through 14, 2025 drew 1.5 million learners, the largest AI course cohort in history. Five days on agents only using Google ADK plus Gemini plus MCP plus A2A plus Vertex AI Agent Engine. Now self paced on Kaggle Learn Guide.
Kaggle AI Agents: What You’ll Learn
This is the production focused agent course on this site. Where Post 01, the Gen AI Intensive, gives you broad literacy, this course gives you production depth on the Google Cloud agent stack. Five days, ten codelabs, one capstone. Build agents with the Google Agent Development Kit, ground them with Model Context Protocol, connect them to other agents with Agent2Agent Protocol, deploy to Vertex AI Agent Engine.
Kaggle AI Agents: The Five Day Agenda
- Day 1, Introduction to Agents. Whitepaper on agent taxonomy, levels 0 to 4, Agent Ops discipline. Codelabs build first ADK agent with Google Search, then first multi agent system.
- Day 2, Agent Tools and MCP. Whitepaper on MCP architecture and enterprise readiness. Codelabs on turning Python functions into agent actions and MCP plus long running operations with human approval.
- Day 3, Context Engineering. Sessions versus memory. Stateful agent via session and working memory. Long term memory persisting across sessions.
- Day 4, Agent Quality. Observability with logs, traces, metrics. Feedback loops with LLM as a Judge and Human in the Loop.
- Day 5, Prototype to Production. Deployment, scaling, A2A Protocol. Codelabs on multi agent systems with A2A and deploy local agent to Vertex AI Agent Engine.
Prerequisites
The Gen AI Intensive on this site is the recommended prerequisite. Kaggle positions the Agents Intensive explicitly as the follow on. Python is highly recommended. A free Kaggle account with phone verification. A free Google AI Studio account for the Gemini API key.
The Google ADK Stack
Google Agent Development Kit is the primary framework. Gemini is the model. MCP is the tool protocol. A2A is the agent to agent protocol. Vertex AI Agent Engine is the deployment target. Notably no LangChain or LlamaIndex in the course, it is Google stack native.
Capstone Project
The capstone launched November 14, 2025. Build and deploy an end to end AI agent system. Submission via Kaggle Notebook. Prizes plus Kaggle social feature for top entries. Public capstone examples include SupportPilot for IT support, LearnLytics 2.0 for student behavior analytics, SENTINELS for disaster intelligence.
A Worked Example, Day 2 MCP in Depth
Day 2 is the MCP day. The codelab walks the same MCP primitives the Anthropic Academy MCP course covers, but using Google ADK as the host. You turn a Python function into an agent action, then expose the same function as an MCP server, then connect the agent to it via MCP instead of direct tool declaration. The pattern matters because MCP is the interoperability layer that lets your agent use tools built by other teams without code changes.
Kaggle AI Agents: Common Mistakes to Avoid
- Skipping the Gen AI Intensive first. Kaggle itself positions Gen AI Intensive as the prerequisite.
- Expecting LangChain or LlamaIndex coverage. The course is Google stack native. Do the HF course on this site for framework comparison.
- Treating ADK as production ready in early November 2025. Participants reported Gemini itself was not fully trained on ADK at launch. Use the codelab code as truth, not AI coding assistants.
- Missing the Vibe Coding June 2026 sequel. A sequel course ran June 15 to 19, 2026 with updated Gemini 3 and Google Antigravity.
Kaggle AI Agents: Best Practices
- Do the Gen AI Intensive on this site first.
- Fork each codelab into your own Kaggle account on the day you start it.
- Pair with the Microsoft AI Agents course for cross vendor comparison.
- Use the ADK code from the official codelabs as truth, not AI coding assistants.
- Submit a capstone if you want the badge with distinction.
Kaggle AI Agents: Frequently Asked Questions
Is the Kaggle AI Agents Intensive free?
Yes. Self paced on Kaggle Learn Guide. All whitepapers, podcasts, codelabs free.
Do I need a Kaggle account?
Yes for codelabs, with phone verification. Plus a Google AI Studio account for the API key.
How is this different from the Gen AI Intensive?
Gen AI is broad survey across five topics, one of which is agents. This is five days all on agents.
Does it issue a certificate?
Kaggle digital badge. Capstone with prizes. Not a university credential.
What framework does it use?
Google Agent Development Kit, Google ADK. Not LangChain or LlamaIndex.
Is there a 2026 sequel?
Yes. The Vibe Coding edition ran June 15 to 19, 2026.
Kaggle AI Agents Intensive is the production focused sequel to Gen AI Intensive. Pair with the HF and Microsoft courses for cross vendor comparison.