PowerMind Orchestrate Overview
PowerMind Orchestrate lives in PowerMind AI Studio under DataOps. You build a pipeline, run it on a spreadsheet (CSV), then score new rows with Predict.
Always pick the right project in the top bar (example: AI Project). Screenshots: Web UI. First walkthrough: Getting Started.
Why this exists
Without a tool like this, people train models in notebooks, lose the file, and cannot show who ran what.
Orchestrate keeps that in one place: create the job in Studio, store the files, run the steps, score new data, and keep a history.
It is for tables with a yes/no (0/1) target, not for chat over documents or large Spark jobs.
What it is good for
Typical jobs: claim risk, churn, loan default, simple fraud flags — anything you can put in a CSV with a label column.
- Train on history, then Predict on new rows.
- Repeat the same steps: Ingest → Train → Validate → Deploy → Monitor.
- See who ran a job (Audit Logs) and how the platform is doing (Metrics).
How it fits together
You work in Studio. Studio talks to the Orchestrate service. Behind the scenes that service uses a scheduler (Airflow), file storage, and optional experiment tracking.
You fill in a form to create a pipeline (no YAML in the UI).
How a job runs
- Create — name the pipeline and tick stages (later stages need the earlier ones). Status Registered means it is ready. Wait a minute the first time so the scheduler can see it.
- Run — upload a CSV that includes
label(0 or 1). Other columns are inputs. - Wait — Run History until Success. Files appear under Artifacts (including the saved model).
- Predict — paste new rows as JSON without
label. Column names must match the CSV. Ignore leftover loan examples in the editor if your data is different.
Sign in
Use your organization account. Studio opens Orchestrate with your authenticated session.
Authorized integrations use credentials issued by a PowerMind administrator. See the API Reference.
What “done” looks like
| Check | Where |
|---|---|
| Healthy | Orchestrate header |
| Registered | Pipelines |
| Success | Run History |
| Model file present | Artifacts |
| Scores | Predict |
Your pipelines and files are yours. The tool must be enabled on the project. Perfect scores on a tiny CSV usually mean the data was too easy — not a production proof.