Creating a space

1. Creating a Genie space

You've used Genie as a consumer. Now let's shift perspective and set up a space from scratch.

2. When to create a space

A Genie space is a curated environment for a specific business domain. At our Bakehouse example, you might create one for franchise sales analytics, supplier performance, inventory, or regional performance. Each space connects to relevant tables and is configured to understand that domain's terminology and common questions. The key principle is focus: each space should serve a clear purpose rather than trying to answer every possible question about your entire data warehouse.

3. Prerequisites

The minimum viable space needs four things: data registered in Unity Catalog, table descriptions, meaningful column comments, and an active SQL Warehouse. But how does data get into Unity Catalog in the first place? Your data team might upload CSV files, connect to external data sources, or register existing warehouse tables. That's typically handled before you create a Genie space. Genie does not discover data on its own; it only knows what you explicitly invite into the room. The most important prerequisite is data quality. If Unity Catalog tables have cryptic column names like col_ax_19 and no descriptions, Genie will be born confused.

4. Space creation in three steps

The setup wizard is simple: just three steps. Click Genie in the sidebar, then New in the upper-right corner. Browse Unity Catalog, select the data tables relevant to your space, and click Create. Behind the scenes, three components come together: the SQL Warehouse runs the code, Unity Catalog provides the library of table names and metadata, and the Genie space is the chat window where it all connects.

5. Example

Browse Unity Catalog and select tables relevant to Bakehouse franchise analytics: sales_transactions, sales_suppliers. Genie is only as smart as the boundaries you set. A few relevant tables beats fifty random tables every time. Too much data leads to ambiguous join paths and hallucinations. Use the Sample Data button to preview before adding.

6. Reviewing suggested queries

After adding tables, Genie automatically searches for popular workspace queries associated with those assets. If relevant queries are found, a notification appears in the Data tab of the Instructions panel. You can review each suggestion: accept the ones that match common business questions, and reject the rest. Accepted queries become example SQL queries in your space, giving Genie a head start before you even write your own.

7. Testing your new space

With tables added and suggested queries reviewed, run a test query to verify everything is connected. Start with a simple question like "Show me total sales by franchise." This tests whether Genie can find the right tables, understand the relationships, and return meaningful results. If this works, the foundational setup is correct and your Bakehouse space is ready for franchise owners.

8. Common mistakes

Three common mistakes when creating spaces. First, the Data Dump: adding fifty-plus tables just in case creates complexity. Leads to ambiguous join paths and hallucinations. Second, granularity mismatch: including both Daily Sales Summary and transaction-level tables. Without instructions, Genie may double-count. Third, the Ghost Warehouse: selecting a SQL Warehouse that is under-provisioned or has aggressive auto-stop leads to frustrating delays.

9. Let's practice!

Explore a Bakehouse space and see how the key components come together. Let's practice!

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.