Skip to content

Add context

Teach Terno AI about your data with descriptions, memories, and example questions so its answers are more accurate.

Terno reads your table and column names, but names don’t always tell the whole story. A little context makes answers much more accurate, especially with many tables or unclear names.

When you add a database, leave Start exploring my database ticked. Terno opens a background chat that studies the data and learns its tables, joins, and metrics. You’ll find the chat in your history.

Fill in the Description when you add a datasource, for example:

Orders, customers, and payments for our online store. Revenue is in USD and excludes tax. Test orders have is_test = true.

In Admin → Terno DBI, open Tables and Table columns to add descriptions. Admins can also ask Terno to draft descriptions for every table and column from the Data sources list.

Good descriptions explain:

  • What a table holds, such as one row per order line.
  • Which table to use, such as use product_catalog for product names.
  • How tables join, such as orders.customer_id = customers.id.
  • Business rules, such as only count orders with status completed as revenue.

Memories are facts Terno uses in every chat, such as your fiscal year start, what active customer means, or which table is the source of truth. Add and edit them from Memory in the profile menu. With DBI Admin turned on in the + menu, admins can also ask Terno in a chat to save a memory for the whole workspace.

In Admin → Terno DBI → Prompt examples, save questions with the correct SQL. Terno uses them as a guide for similar questions.

Tell Terno about your role and business in Personalise Terno, so its suggestions and answers fit your work.