Six dimensions. Eighteen points. One arc from raw data platform to production-grade AI infrastructure. Every point is a concrete, measurable property of your data estate.
Data platform can support human workflows but AI agents will fail silently: wrong answers, unauditable outputs, no quality gates.
Selective AI use cases are viable. Specific dimensions are blocking production; targeted fixes get you to the next capability tier.
AI agents can reason correctly, produce auditable outputs, operate within governance boundaries, and be evaluated for correctness.
Each is scored None, Partial, Mostly, or Complete. The six are independent, and a weakness in any one undermines the others, because AI systems expose all six failure modes at once.
Column-level descriptions in INFORMATION_SCHEMA, kept current. The model cannot reason correctly about data it cannot describe.
MetricFlow semantic models and metric definitions. Without semantics, AI generates plausible-but-wrong answers to business questions.
dbt data_tests: or equivalent quality assertions. Quality checks have to gate AI inference up front; monitoring after the fact is too late.
Golden datasets, snapshot history, regression harnesses. AI outputs cannot be verified without reference data to compare against.
Column masking policies, row access policies, audit logging. AI agent queries must be governed with the same rigor as human queries.
Upstream and downstream lineage queryable at column level. Without lineage, AI outputs are unauditable and incidents are undiagnosable.
OpsPulse starts at 4/18, a realistic baseline for most enterprise data platforms. Each chapter closes one or more dimension gaps; the score is the state of the platform at the end of that chapter.
| dimension | ch1 | ch2 | ch3 | ch4 | ch5 | ch6 | ch7 | ch8 | ch9 | ch10 | ch11 | ch12 | ch13 | ch14 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| D1 Schema Freshness | 1 | 2 | 2 | 2 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 |
| D2 Semantic Coverage | 0 | 1 | 1 | 2 | 2 | 2 | 2 | 3 | 3 | 3 | 3 | 3 | 3 | 3 |
| D3 Test Coverage | 1 | 1 | 1 | 2 | 2 | 2 | 2 | 2 | 3 | 3 | 3 | 3 | 3 | 3 |
| D4 Eval Readiness | 1 | 1 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 3 | 3 | 3 | 3 |
| D5 Governance Scope | 1 | 1 | 1 | 1 | 1 | 1 | 2 | 2 | 2 | 2 | 2 | 3 | 3 | 3 |
| D6 Lineage Coverage | 0 | 2 | 2 | 2 | 2 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 | 3 |
| total | 4 | 8 | 9 | 11 | 12 | 13 | 14 | 15 | 16 | 16 | 17 | 18 | 18 | 18 |
Select the level (0–3) that best describes your platform for each dimension. Same 18-point checklist as Chapter 1 of the book.
Each chapter closes specific dimension gaps in a deliberate order. The OpsPulse arc takes you from wherever you are to 18/18, one concrete pattern at a time.