Tutorials

Getting-started gets you a first result with your data. Tutorials go further: complete scenarios where the right answer is known, plus short recipes for single jobs.

Full scenarios

• Renewal risk on sample CRM data — the canonical first scenario: connect, predict churn risk, defend the forecast from evidence.
• Triage support tickets — classify threads, route by topic, deliver the queue with approval.
• Forecast demand — numeric forecast from ontology state with a measured outcome.
• Clean up duplicates — merge entities across two systems with evidence for every merge.

Short recipes

• Set up a data endpoint — name a door, point it, prove it works.
• Pipeline from Postgres to S3 — nightly extract with an approval-gated write.
• Gate a write with approval — node-level human gate in six steps.

Image: scenario card wall with definitions of done.