Blog
Notes on AI data readiness
Data governance for AI, compliance deadlines that land on your data team, and an honest take on what a readiness score can and cannot tell you.
Is your PostgreSQL data AI-ready? The catalog already knows.
Postgres was never asked to describe itself, only to serve an application. That is exactly the gap an AI workload falls into. Scoring the estate from catalog metadata, without reading a row.
Is your Databricks data AI-ready? Unity Catalog can already tell you.
Databricks is where the AI workload will actually run, which makes its readiness question the least abstract of any platform. Scoring the estate from Unity Catalog metadata, without reading a row.
AI readiness is not data quality.
Pristine values with no documentation pass every quality check and fail every AI workload. What the two scores measure, why competent teams have exactly this gap, and which layer to measure first.
Your warehouse scored 52/100. Here is the order to fix it in.
Teams fix readiness findings easy-first: documentation sprints while the unmasked PII waits a quarter. The remediation order that works, and why the re-scan after each batch is the receipt.
Find the unmasked PII before your AI does.
An AI workload reads everything its role can see, at scale, forever. How a metadata-only scan builds a PII exposure inventory as a set difference, days before Article 10 becomes enforceable.
Is your BigQuery data AI-ready? Auditing the sprawl.
BigQuery estates grow by team, not by design, and the ungoverned tail is where the exposure lives. Scoring readiness from INFORMATION_SCHEMA, policy tags, and labels, with two IAM roles and zero row access.
Is your Snowflake data AI-ready? The metadata already knows.
Snowflake ships the best governance surface in the industry and the median account uses a fraction of it. What ACCOUNT_USAGE, tags, and masking policies reveal about readiness, without reading a row.
The AI data readiness checklist: 12 checks before you build.
Most readiness checklists are aspirations wearing a checkbox costume. Twelve concrete checks, every one verifiable from warehouse metadata, grouped by the six factors that actually fail.
AI readiness is not a one-time score: monitoring data drift.
A readiness score is a photograph of a moving thing. Why the number decays as data drifts, how scheduled metadata-only re-scans catch it, and the materiality gate that means you only hear about it when something real changes.
How MortarIQ scores AI readiness without reading your data.
The thing that stalls most readiness checks is the security review. Why a metadata-only assessment reads the structure around your data and never a single row, what it can and cannot see, and how the connection is locked down.
EU AI Act Article 10: the data governance deadline is August 2, 2026.
The high-risk obligations become enforceable in about seven weeks, and Article 10 is aimed at your data team, not your legal team. What it requires, why it cannot be retrofitted as paperwork, and what to do in the time left.
Is your data ready for AI? Now you can know in minutes.
Most AI projects stall on data, not models. Why we built a readiness scanner that scores your warehouse across six factors from metadata alone, and what a real scan finds.