Skip to main content
Simon Sangla
All work

Audit & AI readiness

Snowflake & dbt Auditor

Audits the dbt models in a Snowflake schema, checks which business questions are safe to hand to AI (the Trust Sprint toolkit's report), and reads the Cortex account-audit notebook's findings. The public demo replays real runs recorded on the public dbt jaffle-shop project, with two defects seeded on purpose, so viewing it runs no query.

Next.js · Snowflake · dbt manifest · Trust Sprint toolkit

Evaluation example

Can your AI analytics answers be trusted?

Reproducible demonstration on the public dbt jaffle-shop project. Two defects were intentionally seeded. This is not a customer engagement or a measured client outcome.

Snowflake & dbt Auditor

The buyer's decision

Before rolling out Cortex or text-to-SQL, an analytics leader needs to know which questions are governed, where model health breaks, and what must be fixed first.

What the example examines

The Auditor replays recorded Snowflake runs to inspect dbt model health and question readiness. A separate Cortex account-audit notebook's findings can be read locally in the browser.

What the evidence shows

The recorded example includes two intentionally seeded defects and a question-readiness verdict. The demo is a replay: opening it does not query a customer system.

Decision this supports

Use the findings to decide whether to repair definitions and rerun validation or move to a bounded sandbox evaluation. This is a proposed decision path, not a client result.

The paid engagement

AI Analytics Trust Sprint: five delivery days, €4,500 HT, 50% on order and 50% on delivery. Production access is read-only; the governed semantic view is built in a sandbox or clone. Deliverables: metric trust map, semantic view, validation harness and readiness report with a go/no-go recommendation.

The sample readiness report shows the format of the sprint's report on fictional demo data. It is not a client outcome. View the sample report

Discuss a Trust Sprint (opens in new tab)