Build an AI Agent knowledge base using SQL (BigQuery + Gemini)

GCP credit → https://goo.gle/handson-ep2-lab1
Codelab & source code → https://goo.gle/scholar
ML in BigQuery → https://goo.gle/3O5squw

Did you know you can call a Gemini model directly from a SQL query in BigQuery?

In this hands-on codelab, Ayo and Annie do exactly that, and use it to solve a real problem: converting messy, unstructured text into clean, structured data at scale.

This is Episode 1 of our multi-part series where we build a fully functional, data-aware AI agent on Google Cloud.

🛠️ *What we cover:*

• Loading raw text files from Cloud Storage as BigQuery external tables
• Using BQML.GENERATE_TEXT to send prompts to Gemini inside SQL
• Parsing and structuring LLM output using JSON functions in BigQuery
• Building a clean, queryable dataset ready for downstream AI pipelines This pattern is incredibly powerful for any team sitting on a mountain of unstructured documents, and wanting to make them queryable without a heavy ETL pipeline.

Chapters:

0:00 - Intro

1:44 - Claim GCP credit

2:40 - Data project overview

4:31 - Project set up

15:00 - ELT extraction loading transform intro

18:09 - Loading data

26:24 - BigQuery external table

33:52 [BQML] ML Generate In BigQuery

Watch more Hand on AI → https://goo.gle/HowToWithGemini
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech

#Gemini #GoogleCloud

Speakers: Ayo Adedeji, Annie Wang
Products Mentioned: Gemini, BigQuery