Unified incident list
Stale tables, failed models/tests, and cost spikes share one prioritized queue you can scan in under a minute.
Use case · Reliability
Watchdog / Fix first
AI ranks stale tables, failed dbt runs, and cost spikes into one Fix-first queue.
The problem
Freshness, dbt, and cost each fire in different channels. Nobody knows what to fix first. The morning standup becomes archaeology instead of decisions.
The outcome
VectorData’s AI Watchdog ranks incidents by blast radius, groups related failures, and hands you a short Fix-first list — metadata only, no warehouse row reads.
1 queue
Freshness · dbt · cost in one place
Root cause
Grouped — not a page storm
Metadata
INFORMATION_SCHEMA + job stats + artifacts
Capabilities
Fix-first ranking
Blast radius and severity decide order — leadership KPIs before unused staging tables.
Calm status, not noise
Warning vs critical states you can explain in standup without opening five tools.
Demo before you connect
Walk the broken-run scenario in /app with sample warehouse metadata.
- ✓Same command center as freshness, dbt failures, and cost spikes
- ✓Reads BigQuery metadata and dbt run artifacts — never result rows
- ✓Plans from $99/mo; Pro includes the full reliability surface
- ✓Open the live product demo without a service account
Related
Executive KPI trust
Know the KPI tables are fresh before the Monday leadership meeting.
Read →Board deck confidence
Freeze the numbers that go in the board pack — with a freshness check first.
Read →Data contracts / SLOs
Freshness targets and owner expectations per table — on the roadmap.
Read →Pipeline on-call
Escalation and mute windows for data incidents — without pager chaos.
Read →