Agentic Data Cloud & Knowledge Catalog Demo¶
Overview¶
The Agentic Data Cloud Demo demonstrates how to build an AI-ready data cloud infrastructure on Google Cloud and perform advanced data analytics using Google Cloud Dataplex, BigQuery, and Gemini models.
In this lab, you interact directly with live eCommerce data in BigQuery and run automated metadata pipelines in Google Colab Enterprise:
- Automated Data Profiling & Quality: Execute Dataplex DataScans to profile tables and validate data quality rules automatically.
- Automated Column Insights: Generate intelligent column descriptions and documentation across dataset tables.
- Relational Business Glossary: Import structured business taxonomies and link glossary terms to physical BigQuery schemas.
- BigQuery Property Graph Analysis: Construct property graphs and query multi-hop customer relationships natively in SQL using GQL syntax.
- BigQuery Generative AI & Vector Search: Connect remote Gemini models to BigQuery and generate embeddings for semantic similarity search.
- BigQuery SQL AI Functions: Run high-level declarative AI functions
(
AI.CLASSIFY,AI.SIMILARITY,AI.IF,AI.SEARCH, and Distillation) directly inside SQL.
What You Need¶
To complete this lab:
- Use a modern internet browser (Google Chrome recommended).
- Note the lab Duration in Qwiklabs. Plan your schedule to complete all steps within the allotted time. When you start the lab, you cannot pause and resume later.
- Do not use your personal Google Cloud account. The lab provides a dedicated Google Cloud project and temporary student credentials.
- Open a new Incognito window in Google Chrome for the lab session to prevent account conflicts.
Start Your Lab¶
Click Start Lab in the Qwiklabs panel. The lab environment provisions foundational Google Cloud resources automatically:
- An isolated Google Cloud project with all required BigQuery, Dataplex, and Vertex AI APIs enabled.
- The
thelook_ecommerceBigQuery dataset populated with eCommerce tables (distribution_centers,events,inventory_items,order_items,orders,products,users). - A dedicated Virtual Private Cloud (VPC) network (
adc-demo-vpc) and private subnet. - A pre-configured Colab Enterprise Runtime Template attached to the VPC.
- A Google Cloud Storage bucket containing all 7 interactive lab notebooks and glossary resource files.
Sign In to Google Cloud Console¶
Locate Your Student Credentials¶
Locate the Connection Details panel in Qwiklabs. Copy the temporary Username and Password.
Access the Console in Incognito Mode¶
- Open a new Incognito window in your browser.
- Navigate to the Google Cloud Console.
- Enter the assigned student Username and click Next.
- Enter the temporary Password and click Next.
- Accept the terms and conditions. Do not add recovery options or sign up for free trials.
Step 1: Review Student Visible Outputs¶
Return to the Qwiklabs interface and inspect the Student Visible Outputs section. Note the following values:
- Colab Runtime Template ID: The pre-configured Colab Enterprise template name.
- Notebooks GCS Bucket: The Cloud Storage bucket containing the lab notebooks.
- Google Cloud Project ID: Your assigned temporary Google Cloud Project ID.
- Google Cloud Region: The primary deployment region (for example,
us-central1). - BigQuery Dataset ID: The pre-populated eCommerce dataset
(
thelook_ecommerce).
Note
All notebooks are provided in both English (base filename) and Korean (_ko
suffix). Choose your preferred language when importing.
Step 2: Open Colab Enterprise¶
- In the Google Cloud Console search bar, search for Colab Enterprise and select it from the results.
- In the top-right toolbar, verify that the Region dropdown matches your
assigned Google Cloud Region from the outputs (for example,
us-central1). - In the left navigation menu, click My Notebooks.
Step 3: Import Lab Notebooks from Cloud Storage¶
Import the interactive lab notebooks directly from Cloud Storage using the import dialog:
- Click Import notebook in the Colab Enterprise toolbar.
- In the import modal, select Google Cloud Storage as the source.
- Click Browse to open the Cloud Storage file selector.
- Select your assigned Notebooks GCS Bucket from the list.
- Select the desired Jupyter notebook file inside the bucket (for example,
01_data_profile_quality.ipynbfor English, or01_data_profile_quality_ko.ipynbfor Korean). - Click Select, then click Import.
- Repeat this process for the notebooks you want to explore during your lab session.
Step 4: Connect to the Pre-Configured Runtime Template¶
- Open the imported notebook in the Colab Enterprise editor.
- In the top-right corner of the notebook editor, click the connection arrow next to Connect and choose Connect to a runtime template.
- Select the pre-provisioned runtime template matching your Colab Runtime
Template ID (for example,
adc-demo-template-...). - Wait for the runtime to allocate and connect.
Step 5: Execute Analytics Exercises¶
Run the notebooks sequentially:
Notebook 01: Data Profile & Data Quality¶
- Open
01_data_profile_quality.ipynb. - Step through the cells to configure Dataplex DataScan jobs.
- Run automated profiling against the
usersandorderstables. - Query profiling and rule validation results directly from BigQuery.
Notebook 02 & 03: Automated Metadata Insights¶
- Open
02_data_insight.ipynband03_dataset_insights.ipynb. - Generate AI-assisted column descriptions using Dataplex DataScans.
- Publish documentation labels into Dataplex Knowledge Catalog.
Notebook 04: Relational Business Glossary Setup¶
- Open
04_glossary_setup.ipynb. - Load the structured retail glossary taxonomy from
resources/business_glossary.json. - Link glossary business terms and synonyms to physical BigQuery tables and columns.
Notebook 05: BigQuery Graph Analysis (GQL)¶
- Open
05_graph_analysis.ipynb. - Create a Property Graph over eCommerce customers, orders, products, and distribution centers.
- Run SQL queries using GQL graph pattern matching
(
MATCH (c:Customer)-[:ordered]->(p:Product)) to identify purchase clusters.
Notebook 06 & 07: BigQuery AI & Remote Gemini Models¶
- Open
06_bigquery_ai_ml_demo.ipynband07_bigquery_ai_functions.ipynb. - Use the pre-configured
vertex-connectionBigQuery connection. - Invoke Gemini models directly in SQL to classify customer reviews, compute semantic vector embeddings, and run cosine similarity search.
- Test high-level SQL AI functions (
AI.CLASSIFY,AI.SIMILARITY,AI.SEARCH).
Step 6: End Your Lab¶
When you finish running your experiments:
- Return to the Qwiklabs browser tab.
- Click End Lab and confirm.
- The lab environment tears down all temporary resources automatically.