// the find
ankane/blazer
Business intelligence made simple
Blazer is a Rails engine for ad-hoc SQL exploration, dashboards, and scheduled checks, mounted directly into an existing Rails app. It's aimed at teams that already have a Postgres/MySQL/Redshift-style warehouse and want a lightweight BI layer without running a separate service like Metabase or Superset.
The adapter system covers an unusually long list of backends (Athena, BigQuery, Snowflake, ClickHouse, Elasticsearch, Cassandra, Presto, Salesforce, and more), and writing a new adapter is a small, documented subclass. Checks with Slack/email alerting give you a basic data-quality monitor for free instead of a separate cron job and notifier. Smart variables and smart columns turn raw foreign key values into human-readable selects/labels without extra joins, which is the kind of detail that makes non-engineers actually use the tool.
Query permissions are just naming conventions (# for unlisted, * for creator-only edits) rather than real ACLs, so anything beyond a small trusted team needs custom authorization bolted on. It only protects against mutating queries by wrapping them in a transaction and rolling back — that's not a substitute for a read-only DB user, and the docs admit as much, meaning it's easy to skip the safer setup. Async execution requires a shared cache store (Memcached/Redis) the moment you run more than one web process, which is an infra requirement most people won't notice until queries start timing out in production. Forecasting and anomaly detection aren't built in — they're thin wrappers around separate gems, and two of the three options (Trend) need a standalone Python service running alongside Rails.