Skip to content

Before you start

Log in

  1. Open the workspace URL your trainer gives you.
  2. Sign in with the account you received. Change the temporary password if asked.
  3. In the left navigation, confirm you can see SQL Editor, Genie, Agents and Playground. If any is missing, tell the trainer.

Set up your data

Nobody has prepared any data for you — you create your own with one click. Nothing to download or clone by hand; one notebook import does it.

  1. In the left navigation, open Workspace. Next to Create in the top right, click ⋮ and choose Import.

    The Workspace page with the ⋮ menu open and Import highlighted

  2. Select URL and paste:

    https://raw.githubusercontent.com/sammydeprez/databricks-agents-training/develop/notebooks/lab_data.py

    Import dialog with URL selected and the raw GitHub link pasted in

  3. Click Import, open the notebook it creates (called lab_data), leave the setting on Set up my data, and click Run all at the top. It takes under a minute. Leave Catalog empty unless your trainer gives you a catalog name — then type it there.

  4. At the bottom, note the catalog and schema it prints, for example enec_training.jsmith. From here on, wherever this guide says your_catalog.your_schema, use that; wherever it says your-name, use the dashed version it also prints (for example jsmith).

    lab_data notebook output confirming the schema, tables and documents were created

Keep this notebook — you'll use it again at the end of Day 2 to clean up, and it's safe to re-run at any point in between too, for example if something looks broken. It replaces your data rather than failing because it already exists.

Your schema now contains:

Table Rows Used in
work_orders 2,000 Lab 2, Lab 5
assets 120 Lab 2
technicians 40 Lab 2
it_tickets 500 Lab 3
incident_reports 60 Trainer demo
meeting_transcripts 10 Lab 3, capstone
monthly_kpis 48 Capstone

And a volume docs with the public regulation PDFs used in Lab 4, plus a memo_templates subfolder used by the capstone, both copied in automatically.

Vocabulary you will hear today

Term Plain meaning
LLM The model that reads and writes text
Prompt What you type to the model, including instructions
Token A piece of a word; models charge and limit by tokens
RAG The model looks things up in your documents before answering
Agent A model that can decide to use tools (run SQL, search documents)
Unity Catalog Where tables, files, functions and agents live, with permissions

Conventions in this guide

  • Bold menu names are things to click, in the order written: Configure > Instructions.
  • Code blocks have a copy button in the top right corner.
  • Replace your_catalog, your_schema and your-name with the names the setup notebook printed for you, everywhere.