Connect, shape, run, ask.
Four steps from a file to an answer you can trace. The agent does the drafting. You decide what is trusted at every step.
Step 01
Connect
Drop files into chat or add a source node. Ordiva profiles each source as it arrives, so every later step starts from real columns.
What you inspect
- File type and adapter
- Column names and types
- Row preview
Step 02
Shape
Describe the outcome. The agent proposes joins, checks, and transforms as nodes on the graph, each with the script it will run.
What you inspect
- Each proposed node
- The instruction behind it
- The generated Python and SQL
Step 03
Run
Execute a node, a branch, or the full pipeline. New scripts pause for review, then each node writes its artifact and schema.
What you inspect
- Run status per node
- Logs and row counts
- Artifacts and output columns
Step 04
Ask
Ask questions against the outputs the run saved. The answer stays tied to the artifact, its schema, and its lineage.
What you inspect
- The artifact it read
- The schema of the result
- The lineage upstream
Ready to build a pipeline you can inspect?
Describe the outcome, inspect what the agent proposes, and run it when you trust it.