Kuunda Sage (data analyst)

What Sage looks at

Sage starts from your actual schema: tables and their approximate volumes, columns and their types, and the state of RLS policies. Its answers come from your database, not from a generic example.

• Volumes: which tables carry rows, which ones stay empty.
• Schema quality: columns never filled, tables with no primary key, indexes your filters often call for.
• Security: state of RLS policies, tables exposed without a policy.
• Modelling: how to split a misplaced field or shape a new table.

How it works

For every figure it states, Sage runs a SELECT query on the project schema, then translates the result into business language: rounded numbers, comparisons, recommendations. It does not invent data and does not print SQL blocks unless you ask for them.

• A single SELECT (or WITH) query per step, limited to 100 rows by default.
• No writes: statements that change data or schema are refused.
• Scope strictly limited to your project schema.
• When a question falls outside SQL, Sage says so and suggests another angle.

Asking a good question

Name the topic and the intent: Sage restates and verifies before answering. A few questions that work well:

• What are my 5 largest tables and how are they growing?
• Do I have tables without a primary key?
• Summarise the RLS state on my schema.
• Which indexes might be missing for my frequent filters?

Quota and availability

Sage is consumed in messages, per project: the counter at the top of its screen shows what is left for the period, and the quota resets every month. When the feature is not enabled on the platform, the screen says so and invites you to contact support.

Next steps

• Policies (RLS)
• Views, triggers and routines
• Usage and quotas