SQL, databases, and agents in plain words

Nobody is born knowing what a query is

This course assumes no SQL and no data background, so it starts by defining its words. Every lesson after this one leans on the six terms below. If you have written SQL before, skim the last section and the checkpoint and move on.

Where data lives

Every app you use writes records somewhere. An order placed, a refund issued, a signup completed - each becomes a row in a database, which is a program whose whole job is to store records and answer questions about them.

Inside a database, records are organised into tables. A table is a grid: the columns name what is stored, and each row is one record. Here are three rows of an orders table like the one you will work with:

If you have used a spreadsheet, this shape is familiar: named columns across the top, one record per row. The difference is scale and discipline. A database holds millions of rows, many tables, and answers questions about all of them at once.

What SQL is

SQL is the language for asking a database questions. A query is one question, written in SQL. This is a complete query:

SELECT order_id, order_total
FROM orders
WHERE order_total > 500;

Read it as English: from the orders table, show the id and total of every order over 500. You send the query to the database, and it sends back a result - rows and columns, the same shape as a table.

SQL has been the standard way to do this for fifty years. Every company that has data has SQL somewhere, and every tool in this field speaks it. That is why it is worth reading even in the age of AI: it is the language the answers are written in.

What a warehouse is

Companies copy data out of their apps - the sales system, the website, the support tool - into one database set aside for analysis. That database is called a data warehouse. The point is to put everything in one place, so a single query can see all of it.

The warehouse in this course is DuckDB: a real analytical database that lives in a single file on your machine. The ideas are the same ones you would use on Snowflake or BigQuery at a job. Ours is just small enough to run locally, for free, with nothing to sign up for.

What an AI coding agent is

An AI coding agent is not the chat website. It is an AI you run on your own machine - Claude Code, Cursor, or Codex - that can read your project's files, run commands in your terminal, and show you what it did. Where a chatbot can only talk about your data, an agent can actually query it.

One more name you will see everywhere here: Bruin, the command-line tool this course runs on. Command-line means you type commands into a terminal - a window that runs typed commands instead of clicks - and read the text that comes back. Bruin is what runs your queries against DuckDB and builds the project. The setup lesson, two steps ahead, installs all of it step by step, assuming you have never opened a terminal.

Why learn SQL if the agent writes it

Because the agent's answer is a number, and someone has to decide whether the number is right. A query can run without error and still be wrong - and the only person who can catch that is one who can read the query. The next lesson makes this argument properly. For now it is enough to know the course's shape: you learn just enough SQL to read it, then you learn to check SQL you did not write.

Checkpoint

In your own words, no looking back:

• What is a table, and what does one row of it represent?
• What is a query?
• What does a warehouse hold that a single app's database does not?
• What can a coding agent do that a chat website cannot?