Terminology
A comprehensive glossary of terms used throughout the ProAI platform.
| Term | Definition |
|---|---|
| Pipeline | A complete data flow from source(s) through transformations to destination(s) |
| DAG | Directed Acyclic Graph — the execution graph for a pipeline |
| Node | A single unit of work within a pipeline (source, transform, or destination) |
| Edge | A connection between two nodes representing data flow dependency |
| Connector | A pluggable adapter for external system integration |
| Job | A single execution instance of a pipeline |
| Run | An execution attempt of a job, which may be one of several retries |
| Schedule | A cron-based or event-driven trigger for job execution |
| Transform | A data manipulation operation (filter, map, join, aggregate) |
| Watermark | A timestamp tracking the progress of streaming data processing |
| Checkpoint | A saved state allowing pipeline recovery from failure |
| Lineage | The complete history of data transformations from source to destination |
| Agentic Flow | An AI-powered workflow that autonomously optimizes pipeline behavior |
| Canvas | The visual drag-and-drop interface for building pipelines |
| Workspace | The ProAI web application environment |
| Project | A version-controlled collection of pipelines, transforms, and configurations |
| Environment | An isolated deployment target (dev, staging, production) |
| Service Account | A non-human identity for programmatic access |
| Connector Profile | A saved configuration for connecting to a specific data source |
See Also
- Key Concepts — Core mental models
- Architecture — System design