Bigeye

Bigeye is a data observability platform for data teams, with lineage, anomaly detection and YAML or SQL quality rules on cloud warehouses, sold by quote.

Bigeye warns a data team when a table in its warehouse has gone wrong and shows how far the damage spreads, using lineage to identify, triage and resolve data incidents. Kyle Kirwan and Egor Gryaznov founded the company of the same name in 2019, and it remains independent. From that observability base the platform has grown into an enterprise AI trust platform that also covers metadata, data classification and the data AI agents reach. Monitoring starts with metrics, deployed from the interface by choosing a metric type and the columns to watch, with Bigeye suggesting columns. Autothresholds learn each metric's normal range from its history, seasonality and trend, and manual thresholds remain an option. Engineers can instead define monitoring in bigConfig, a YAML configuration applied through the Bigeye CLI, or write Custom SQL Rules and parameterized Templates that run across many tables. Data profiling reports nulls, distinct values, duplicates and recognised patterns for chosen columns, and the Scorecard shows the share of healthy metrics for each source, schema, table and column. Failed metrics become issues with a priority, a status, an assignee and a comment timeline, and related issues group into incidents. Deltas compare two similar datasets, and Agent Trust Hub and AI Guardian monitor and govern how AI agents use enterprise data. The platform is aimed at data engineering and data platform teams in larger organisations whose data lives in Snowflake, BigQuery, Redshift or Databricks. Engineers who prefer configuration files work through bigConfig and the CLI, while analysts and data owners can deploy metrics and follow issues from the interface. It also suits companies moving to a new warehouse, and enterprises whose governance staff need to control how AI agents access company data. A common job is guarding a pipeline: the RunMetricsOperator runs a table's metrics inside an Airflow DAG, and circuit breaker mode stops the run when a metric alerts. Analytics engineers use Deltas to check that a dbt model change or a migrated table still matches the original before it ships. When something breaks, the alert reaches the team in Slack, opens a two-way ticket in Jira or ServiceNow, or pages the on-call engineer through PagerDuty. Data Health then carries each table's status into Alation, Atlan or data.world, so the people reading a dashboard can see whether its source passed its latest checks.