DatronomeFeatures
Everything here ships in Datronome today. The free Starter plan includes approvals and the decision log; paid plans add more pipelines, people and files, plus alerts and server mode. Pricing has the details.
Decide and prove it
This is what makes Datronome different: the pipeline ends with a person making a call, and Datronome keeps the record.
Approvals
A pipeline can pause and ask a named person to approve. They see the real table in their inbox, approve or reject with a comment, and the pipeline carries on or stops. See an example.
Decision log and outcomes
Every decision is recorded with who, when, the data shown and the value at stake. Days later Datronome can check whether it worked, and shows your team's hit rate and cost per decision. See an example.
Receipts
Download any decision as a receipt: what was shown, who decided, their comment and what happened next. It includes a SHA-256 checksum, so a later change to the downloaded file shows up when the checksum is checked again.
Catch problems
Wrong numbers are stopped before anyone acts on them.
Quality checks
Rules in plain lines: rows >= 1, not_null: account_id, unique: invoice_id, range: amount 0 50000. anomaly: amount flags a total that jumps away from its history. A failed rule stops the pipeline. See an example.
Check summaries by email
Every check writes a short report: row count against the usual, columns that moved, blanks that appeared. Have it emailed on failure, when something moves, or every run.
Run it
The everyday work: get the data, shape it, and put it on a schedule.
Pipelines on a schedule
Chain steps on a canvas: SQL, Python, Windows commands, API calls, reading a mailbox, moving files, sending email. Schedules are plain words like "weekdays at 6:30", not cron codes.
Try it before you schedule it
Every step has a Try button that runs just that step on the latest data, with no schedule and no history. A full-screen code editor knows your tables and columns.
Database connections
Short forms for BigQuery, Databricks, Snowflake, Redshift, Postgres, SQL Server, Azure SQL, MySQL and SQLite. Passwords are stored encrypted and never shown again. Browse tables before you write a query.
Dashboards and Explore
Turn any result into charts and tables, filter them, sort any column, and arrange tiles on a dashboard. Dashboards can be emailed on a schedule.
Also included
Predict and Optimize
Train a model on a CSV and score new rows, with an honest hold-out score. Or find the best plan under limits written in plain lines, such as budget: cost <= 250000. See an example.
The assistant
Describe what you need in plain words. The assistant drafts a pipeline, tests it in a scratch folder, and hands you a card to review. Nothing is saved until you press Create. It can use Hugging Face, Anthropic, or a model on your own machine.
People and roles
Owners manage people, editors build and decide, viewers look and download. Every save is a version you can restore, and pipelines can be kept in a GitHub repository.
Where your data lives
Datronome runs on your own computer or server. Your pipelines, files, results and decisions stay there; we never receive them. The only connections out are the ones your steps make (your database, an API you call) and, if you turn them on, email alerts and a cloud assistant, which sees pipeline code, column names and run logs. A paid licence checks in about once a day with its identifier and the version number, and nothing else. The security page has the details, and the Privacy Policy says exactly what we collect.
What it is not
Datronome runs on one machine. It is not a cluster scheduler for hundreds of pipelines, and it has no backfills or file-arrival triggers yet. It does not replace dbt's model graph. The comparisons are honest about where other tools are the better choice.