Request a demo

Analysis

Ask questions of what actually ran.

Ask in plain language. The agent reads the outputs your nodes persisted, and when the answer needs new work it adds a node, runs it, and answers from that node's output.

Request a demoHow it works

The answer lives on the graph

Answers are artifacts

The answer lives on the graph

Asked which customers failed payment more than once, the agent added top5_failed_customers after the churn score, ran it, read its sample, and answered with the five rows it produced.

  • Reads samples of persisted node outputs
  • Adds a node when the answer needs new work
  • The answer node stays on the canvas to rerun

Analysis in detail

  • 01

    Persisted outputs

    Questions read the artifacts your runs saved, so the answer matches what the pipeline produced.

  • 02

    Answers you can rerun

    When an answer needs computation, it becomes a node you can open, rerun, and review.

  • 03

    Schema in view

    Every node shows the columns it emits, so a question names fields that exist.

  • 04

    Lineage on hand

    Open a column to see which nodes produced it and how it changed along the way.

  • 05

    Outputs you can pin

    Turn a result into an output widget and pin it to the project dashboard.

  • 06

    Ask from anywhere

    The command palette takes a question from any screen with Cmd K.

Related capabilities

All capabilities

Ready to build a pipeline you can inspect?

Describe the outcome, inspect what the agent proposes, and run it when you trust it.

Request a demoLog in
How it worksConnect, shape, run, and ask, step by step.API referenceCreate an API key and call Ordiva from your own code.