5-Day AI Agents Intensive Course with Google
by Google & Kaggle
Five whitepapers, hands-on codelabs and a capstone: build production AI agents with Gemini and ADK.
Overview
This is Google and Kaggle's five-day AI agents course, run live from 10-14 November 2025 and now available free and self-paced as a Kaggle Learn Guide. Over 1.5 million people registered, more than five times the preceding Gen AI Intensive which itself held a Guinness World Record, and participants submitted over 11,000 capstone projects. Each day pairs a technical whitepaper written by Google researchers and engineers with an audio podcast summary, a YouTube livestream, and one or two hands-on codelabs run in Kaggle Notebooks against Gemini and Google's Agent Development Kit. Day 1, Introduction to Agents and Agentic Architectures, covers the model-plus-tools-plus-orchestration structure, the think-act-observe loop and a level 0-4 agent taxonomy. Day 2, Agent Tools and Interoperability with MCP, covers tool design, the Model Context Protocol client-server model, JSON tool schemas and security issues including the confused-deputy problem. Day 3, Context Engineering: Sessions and Memory, splits into two codelabs, one on session services and one on long-term memory pipelines. Day 4, Agent Quality, covers observability with OpenTelemetry, then evaluation with rubrics, LLM-as-a-judge and human-in-the-loop, plus a bonus NVIDIA cuDF GPU notebook. Day 5, Prototype to Production, covers CI/CD, evaluation-gated deployment, the A2A protocol and deploying to Vertex AI Agent Engine. Guest speakers came from Google, Cohere, Reified and NVIDIA, and completing the capstone earns a Kaggle badge and certificate.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Course
- Cost
- Free
- Duration
- 5 days as run live; ~20-25 hours self-paced including whitepapers and codelabs
- Provider
- Google & Kaggle
- Hands-on
- Yes — code/exercises
- Certificate
- Available
What You’ll Learn
- ✓Build a first agent with Gemini and the Agent Development Kit from scratch
- ✓Design tool schemas and connect agents to real APIs through the Model Context Protocol
- ✓Implement session state for immediate context and long-term memory for personalisation
- ✓Instrument agents with OpenTelemetry traces, logs and metrics for production observability
- ✓Evaluate agent output with rubrics, LLM-as-a-judge and human-in-the-loop review
- ✓Reason about agent security, including the confused-deputy problem and gateway mitigations
- ✓Deploy a multi-agent system to Vertex AI Agent Engine behind evaluation-gated CI/CD
- ✓Use the A2A protocol to let separately built agents delegate work to each other
Highlights
- •Completely free, and the five whitepapers alone are substantial — they drew two million views during the live run
- •Every concept has a matching Kaggle Notebook codelab, so nothing stays purely theoretical
- •Covers the unglamorous production half — observability, evaluation, CI/CD — that most agent courses skip
- •The capstone earns a Kaggle badge and certificate; over 11,000 entries were submitted in the live cohort
- •Independent write-ups put the real time cost near 20-25 hours, well above what the '5 days' branding implies
- •Days 4 and 5 are genuinely advanced: expect statistical thinking and DevOps concepts, not a gentle beginner ramp
Who It’s For
Best For
- ✓Developers moving from LLM prompting to building real, tool-using agents
- ✓ML engineers who need a fast, structured grounding in agent architecture
- ✓Teams standardising on Gemini, ADK and Vertex AI Agent Engine
Prerequisites
- •Working Python skills for the codelabs, though Day 1's concepts need none
- •Familiarity with HTTP/REST APIs and JSON schemas for the MCP material
- •A free Kaggle account and a Google AI Studio API key
- •DevOps and CI/CD awareness helps considerably on Day 5
FAQ
What is 5-Day AI Agents Intensive Course with Google?
Google and Kaggle's free five-day intensive on building AI agents, now available self-paced. Each day pairs a Google-authored whitepaper with hands-on codelabs on Gemini and the Agent Development Kit, moving from agent architectures through tools and MCP, sessions and memory, observability and evaluation, and finally production deployment with A2A and Vertex AI Agent Engine, ending in a capstone project.
Is 5-Day AI Agents Intensive Course with Google free?
5-Day AI Agents Intensive Course with Google is free to access.
What level is 5-Day AI Agents Intensive Course with Google for?
5-Day AI Agents Intensive Course with Google is aimed at a intermediate audience. Recommended background: Working Python skills for the codelabs, though Day 1's concepts need none, Familiarity with HTTP/REST APIs and JSON schemas for the MCP material, A free Kaggle account and a Google AI Studio API key, DevOps and CI/CD awareness helps considerably on Day 5.
How long does 5-Day AI Agents Intensive Course with Google take?
Expect roughly 5 days as run live; ~20-25 hours self-paced including whitepapers and codelabs. Most learners work through it at their own pace.
What will I learn from 5-Day AI Agents Intensive Course with Google?
You'll learn: Build a first agent with Gemini and the Agent Development Kit from scratch; Design tool schemas and connect agents to real APIs through the Model Context Protocol; Implement session state for immediate context and long-term memory for personalisation; Instrument agents with OpenTelemetry traces, logs and metrics for production observability; Evaluate agent output with rubrics, LLM-as-a-judge and human-in-the-loop review; Reason about agent security, including the confused-deputy problem and gateway mitigations; Deploy a multi-agent system to Vertex AI Agent Engine behind evaluation-gated CI/CD; Use the A2A protocol to let separately built agents delegate work to each other.
Topics
Sources
This page was written from 5 sources, 4 on domains other than kaggle.com.