Timefold Copilot (preview)
|
This guide describes functionality that is currently available as a preview feature. If you’d like early access to this feature, contact us. |
Timefold Copilot lets you interact with your planning models in plain language. Instead of digging through JSON or writing queries, you describe what you want to understand or explore, and the Copilot handles the translation.
Available skills
Explain bottlenecks
Ask the Copilot to analyze a dataset and explain what is limiting your planning solution. You can ask questions like "Why are 12 shifts left unassigned on Friday?" or "Which constraint is being broken most often?", and get a plain-language answer without having to interpret score analysis manually.
The skill also suggests follow-up experiments based on what it finds, so you can move directly from understanding the problem to testing a solution.
Start a what-if experiment
Describe a change to your planning problem in plain language, and the Copilot sets up the scenarios and runs them in parallel using the Timefold Experiments API. You get a side-by-side comparison without having to configure or trigger each dataset manually.
The Copilot supports the following experiment types:
- Goal alignment
-
Verify whether the model is optimizing toward the outcome you actually care about. Useful when you suspect constraint weights or priorities may not reflect real-world needs.
- What-if scenarios
-
Explore how a change in the planning problem affects the outcome. For example: "What happens if workload increases by 20%?", "What if this employee is unavailable all week?", or "What if we add a required skill to this shift category?"