Data pipelines as code for engineering teams
A deploy renames a field. No wrong number goes live.
SQL, Python and YAML, checked on every PR and every load.
Trusted by forward-thinking teams
Next to your service
Pipelines in your repo, shipped through your CI.
/* @bruin
name: stg.checkout_events
type: bq.sql
materialization:
type: table
depends:
- raw.checkout_events
columns:
- name: amount_cents
checks:
- name: not_null
- name: positive
@bruin */
select
event_id,
order_id,
-- v4.12 sends amount in dollars
coalesce(amount_cents, cast(round(amount * 100) as int64))
as amount_cents,
currency
from raw.checkout_events
where event_name = 'order_completed'$ bruin run data/assets/stg_checkout_events.sql --downstream
- stg.checkout_events1.3s
- not_null · amount_cents
- positive · amount_cents
- mart.orders_hourly2.2s
- mart.customer_ltv3.1s
3 assets, 2 checks passed. Nothing downstream is held.
Column-level lineage
Before you rename a field, see what reads it.
Source
- order_id
- amount_cents
- currency
Staging
- order_id
- amount_cents
Mart
- revenue
- ltv_90d
Used by
From a data engineer
“Adding a new product used to be a lot of steps. Now I work in Bruin from VS Code and Cursor, and it is so easy to adapt that we can add a new product fast. It is three or four times faster.”
Mehmet Can ÖzçelikData Engineer, NativeMinds- apps on one platform
- 20+
- faster pipelines
- 3 to 4x
- cheaper than their Fivetran stack
- Up to 5x
Built on the Bruin platform
Product and finance ask. AI answers from your models.
Sources
DatabasesWarehousesApps & APIsFiles & storageStreams & webhooksWeb scrapingMove
Data IngestionFrequently asked
Questions from engineering teams.
Does our engineering team need a separate data team to use Bruin?
No. If your team writes SQL or Python and reviews PRs, it can own its pipelines, starting on a laptop with no extra infrastructure.
How do we get product events and database tables into Bruin?
ingestr loads Postgres and MySQL (read replicas work), Kafka, Segment, PostHog and Amplitude, plus thousands more through APIs. Nothing needs direct access to your production database.
Which data warehouses does Bruin run on?
Snowflake, BigQuery, Databricks, Redshift, Postgres, ClickHouse, DuckDB, MySQL and SQL Server, among others. SQL runs inside your warehouse, so the data stays where it is.
Can we run Bruin pipelines locally on a laptop?
Yes. Install the CLI and run a pipeline on your laptop against DuckDB or a dev schema, with the same checks you get in production. No Bruin account needed.
Can we run Bruin in CI with GitHub Actions?
Yes. Add bruin validate and bruin run to GitHub Actions or any CI, and every pull request builds and checks its models on a CI schema before merge.
Can we run SQL and Python transformations in one Bruin project?
Yes. Python assets sit in the same graph as SQL and ingestr assets, with the same checks and lineage. They can return a dataframe for Bruin to materialize as a table.
What does Bruin do when a deploy changes an event?
Checks on the incoming data catch it, hold everything downstream and post the cause in Slack, so no wrong number goes live. Column-level lineage shows what reads a field before you rename it.
Can Bruin orchestrate our data pipelines without Airflow?
Yes. Dependencies come from the assets, so there are no DAG files. Schedule in Bruin Cloud or CI, and move DAGs over one at a time.
Does Bruin work in VS Code, Cursor or Claude Code?
Yes. The VS Code extension runs assets, previews query results and shows lineage. In Cursor or Claude Code, Bruin MCP lets agents query your data and build pipelines.
How do product and finance get answers from Bruin without pinging engineers?
They ask the AI analyst in Slack, Microsoft Teams, Google Chat, WhatsApp, Discord, Telegram, email or the browser. It answers from your models and tested definitions and shows the query it ran.
How much does Bruin cost, and is it per seat?
The Bruin CLI (open source, Apache 2.0) and ingestr (source-available) are free to run locally or in your CI. On Bruin Cloud, compute is billed per second (about $2.50 an hour for SQL, $10 on a standard instance) and AI tasks cost $1 to $3 each, depending on complexity. Committed-use discounts lower the rate as usage grows. No seats.
Ship the rename. Keep the revenue numbers right.
$100 in credits and 50 AI tasks. No credit card.
Bruin CLI and ingestr are on GitHub.