Gemini 3 with Google Search

This video demonstrates the capabilities of the **Gemini 3 Pro Preview** model, integrated with **Google Search Grounding**, in a VS Code Jupyter Notebook environment. The video follows my real-time process of planning an upcoming trip to Taipei and Taoyuan, starting with the technical setup of the `google-genai` SDK and navigating a live API authentication issue by migrating to Vertex AI. Once the environment is stable, I leverage the model's "thinking" capabilities to retrieve accurate weather forecasts for late November 2025, generate a context-aware packing list based on those conditions, and confirm detailed flight schedules for Cathay Pacific. The session concludes with a manual verification step that successfully validates the AI-generated flight data against live web results. #gemini3 #gemini3pro #GoogleSearch #GenerativeAI #VertexAI #grounding #thinkinglevel | Start | End | Caption | | :--- | :--- | :--- | | 00:00 | 01:33 | Introduction: Planning a trip to Taiwan with Gemini 3 | | 01:33 | 03:33 | Setting up the Environment: Installing libraries and loading API keys | | 03:33 | 05:05 | Explaining helper functions: Token usage and citation building | | 05:05 | 06:11 | Defining the Weather Forecast prompt with Google Search tools | | 06:11 | 08:29 | Troubleshooting: Handling an API error and switching to Vertex AI | | 08:29 | 10:13 | Running the Weather Forecast query for Taipei and Taoyuan | | 10:13 | 13:13 | Reviewing the Weather results, "Thinking" tokens, and citations | | 13:13 | 16:38 | Generating a contextual Packing List based on the itinerary | | 16:38 | 18:18 | Querying for Cathay Pacific flight schedules (CX530 & CX495) | | 18:18 | 21:10 | Reviewing the Flight Information response and data accuracy | | 21:10 | 22:36 | Fact-checking the AI results against actual Google Search results | | 22:36 | 22:44 | Conclusion and summary of the demo | Google Cloud credits are provided for this project.