Engagement · Full Transformation
Full AI Data Transformation
“Get our warehouse into a state where an agent can query it and be right.”
The audit, the rebuild and the context layer run end to end by us: the warehouse read, re-architected into clean layers, given definitions your team agreed, and handed back tested. An agent pointed at an undefined warehouse returns a confident answer; this engagement is about it also being the correct one.
In scope:
- The audit, the rebuild and the context layer delivered as one sequenced engagement rather than three procurements
- Your warehouse re-architected into clean layers, generated and run against your warehouse before handover
- One agreed definition per metric, so a question has one answer regardless of who — or what — is asking it
- Documentation written against the delivered models rather than alongside them
- Scheduled checks on the source tables the rebuild depends on — volume, freshness, schema change, null rates, duplicates, distribution drift
- A working session at each stage, and every artifact yours to keep
What we need
- Read-only warehouse credentials to start; write access to a separate schema at the build stage
- Your dbt project, a folder of SQL, or a Glue catalog
- Named owners for the definitions, and one reviewer with authority to accept the delivered models
How long
Out of scope:
- Deploying into production, orchestration changes and BI rework — the same exclusions as the Medallion Rebuild
- Choosing or standing up your agent platform. We make the warehouse fit to be queried; we do not build the agent
- A guarantee that the warehouse is 'AI-ready' by a fixed date. Readiness is judged against acceptance criteria we agree on the call, in writing, before the first stage starts
What we mean when we say tested.
We generate the models, run them against your Snowflake or Athena connection, and check that each one returns rows, that its primary key is unique, that no column comes back completely empty, and that the types are what we said they'd be. On top of that we run dbt's own generic tests — not-null, unique, accepted values, and foreign keys against their parent — and hand you the schema.yml for them, so the tests we ran are tests you keep. Where a test can't be evaluated soundly it comes back marked skipped, never passed. We don't run a full dbt build, and we don't compare your old numbers against the new ones row by row. If a figure has to reconcile exactly, that's a conversation for the call.
Not here yet
Ships with the beta A single score for how ready your warehouse is for agents is not built yet. Nothing in this engagement rests on it — readiness here means the acceptance criteria we agree in writing before the first stage starts, which is a stricter test than a number.
Written in future tense on purpose. Anything on this list is something we intend to ship, not something you would be buying today.
Scope the engagement in thirty minutes.
Bring the domain that causes the most difficulty. If it turns out this is not the right engagement for you, we will say so and point at the one that is.