AI Prediction node

Status: Current | Last reviewed: 2026-07-07

Calls a trained score model and maps workflow values into its feature inputs.

Behavior

Select a trained Score Model and map workflow values into the model's feature inputs. Each feature row becomes one field on the prediction payload. The selected model supplies the provider, endpoint, and credential integration.

The prediction is written to the node result as score and also to state.vars.<output_variable> for downstream merge templates.

You can add this node from the workflow editor add-node menu when building a flow.

Configuration options

Option Required Type Values / default What it controls
Score model (model_id) Yes string Trained model from the Score Models studio. The model supplies the provider and integration used for prediction.
Features (score_features) Yes array Map each model feature name to a workflow value. Use bulk add to paste many mappings at once.
Prediction variable (output_variable) No string Default: prediction Writes the model prediction to state.vars. so downstream blocks can use {{vars.}}. Advanced setting.
Include provider details (include_breakdown) No boolean Default: true When on, the node result includes confidence, model version, and provider detail when available. Advanced setting.

Outputs

Output Type What it contains
score number Numeric prediction returned by the model.
confidence number Provider-supplied confidence when available.
breakdown object Provider details when include_breakdown is true.
method_used string
model_version string Provider-supplied model version string when available.

See also