Timefold

This plugin integrates Kestra with the Timefold Platform to solve AI-powered optimization problems such as employee scheduling and field service routing.

Prerequisites

Each model on the Timefold Platform requires its own API key. You can generate one from the Timefold Platform. Store it as a Kestra secret (e.g. TIMEFOLD_API_KEY) and reference it with {{ secret('TIMEFOLD_API_KEY') }} in your workflow.

Supported models

Consult the relevant user guide for the exact modelInput schema expected by each model — field names, required properties, and value types differ between them.

Tasks

• Solve: submits a modelInput dataset to a Timefold model and returns a jobId. Optionally polls until solving completes and returns the full modelOutput.
• GetDataset: retrieves the current state of a previously submitted job by jobId. Use this after Solve (with wait: false) to fetch the solution once solving has completed.

Large datasets and long solve times

When solveDuration is omitted, the Timefold Platform uses its built-in diminishing-returns termination to decide how long to run. It stops when further solving is unlikely to meaningfully improve the solution.

For large datasets where solving may take a significant amount of time, set wait to false (or omit it, false is the default) in the Solve task. This returns the jobId immediately without blocking the flow. Use a subsequent GetDataset task to retrieve the solution once solving is complete, polling on your own schedule.