Glossary
What is metadata?
Metadata is data about data: the names of tables and columns, data types, relationships and descriptions. With text-to-SQL, an AI model only needs metadata to write a query, not the rows of data themselves.
Metadata explained
There is technical metadata, such as table names, column names, data types and keys, and business metadata, such as descriptions, units and owners. Technical metadata is usually read automatically from the database; business metadata has to be written down by someone.
With AI on business data, the distinction between metadata and data matters for privacy and confidentiality. A language model that only knows there is a "revenue" column in the "orders" table does not know a single amount. Bear in mind that column names or descriptions can be sensitive too if you put confidential information in them.
The better the metadata, the better the result. A column called "amt2" tells an AI model little; a column called "amount_excl_vat" with a description tells it much more.
Example
The metadata of Acme Groothandel's orders table lists the columns order_id, customer_id, order_date and total_amount, with their data types and the relationship to customers. An AI model can use that to turn "revenue per month" into SQL without seeing a single order.
Metadata in ENABLE
In ENABLE the AI receives only metadata: tables, columns, data types, relationships, the data dictionary and an estimated number of rows per table, plus the user's question. Rows of data and passwords never go to the language model; when explaining or fixing a query, the query and the error message are sent as well.
FAQ
Frequently asked questions
What is the difference between data and metadata?
Data is the information itself, such as the amounts and names in a table. Metadata describes that data, such as the table name, the columns and their data types.
Is metadata personal data?
Usually not: a column name like "customer_name" contains no personal data itself. It can be if descriptions or sample values contain personal data, so be careful about what you record in metadata.
Why does AI only need metadata for text-to-SQL?
To write a query, the model needs to know which tables and columns exist and how they relate. The database then runs the query and supplies the figures, so the model does not need to see the rows.
Further reading
- ResourcesText-to-SQL explained: AI on your databaseText-to-SQL is a technique in which an AI model converts a question in plain language into an SQL query, which is then run on a database. The language model writes the query; the database supplies the figures. How well that works depends mainly on the quality of your metadata and on the safeguards around execution, such as read-only access, timeouts and row limits.Read more
- SolutionSecure AI for business: private AI on your dataWhat private, secure AI really means, and how ENABLE limits what the AI sees of your data.Read more
- PlatformSQL Explorer with AI: ask your database questionsAsk your data warehouse in plain language. The AI writes the SQL, you stay in control.Read more
- SolutionAI on your own data: ask questions of business dataQuestion your data warehouse in plain language. The AI sees the structure, not your rows.Read more
Use AI on your own data. Securely, and without the headache.
In an online demo we show how ENABLE makes your data securely available and how you build on it: describing an app in the AI Builder, plain-language questions, dashboards and APIs. And how we set it up and manage it for you.