Core concepts
Core concepts
The seven ideas that make everything else in predictAI make sense.
predictAI has a small number of core objects. Once you know what each one is and how they connect, every endpoint in the API reads naturally.
| Concept | In one sentence |
|---|---|
| Workspace | The container everything lives in — data, models, deployments, goals. |
| Signal | A named time series: daily_sales, web_traffic, machine_temp. |
| Segment | A curated view over signals — the exact table a model trains on. |
| Model | A trainable artifact: your code, a foundation model, or an upload. |
| Pipeline | A training workflow: segment + model(s) + schedule. |
| Training | One run of a pipeline, producing a scored, deployable result. |
| Deployment | A training promoted to live serving. |
| Goal | An outcome you want forecast — the platform automates the rest. |
Start with the forecasting loop if you want to see how they fit together end to end.