Definitely interested but questions about a few things

[MrKBM]

1. I have spent a couple of hours reading but I am not clear. Do you have to recompile every time you run this or do you pass the parameters.
2. The buzzword AI is mentioned. Does that mean Timefold has the ability to learn?
   I will explain my use case. We provide shop management software to small job shops. X# machines (5-50), X# jobs (20-200 to schedule over 2-3 weeks) each with X# operations (2-50) and a due date. Goal: get all jobs done by their respective due dates.
   Our users are not going to be compiling a program every time they run it but if we can provide an interface and just pass variables to the program form our database that is fine.
   As to the AI. The trick in a small shop of course is there are simply to many tribal knowledge variables for the user to input. However, as an example each job is a part number and while it COULD run on any one of several machines, frequently certain machines are set up for particular parts so they are usually run only on those machines to avoid additional setup or perhaps it is just easier to hold tolerances on some machines. If Timefold has the ability to be fed prior history and weight parts/machines accordingly that would be huge.

[Christopher-Chianelli]

1. No; if you use Python, you don't need to compile at all; if you use Quarkus, it has a developer mode that automatically recompiles Java code for you on change. As a developer, you would either ship a compiled jar that can be run like java -jar path/to/app.jar or a python package.
2. Timefold Solver does not use machine learning. It is a form of classical AI (in particular, a PlanningAI). Unlike machine learning, it uses algorithms written by people. Timefold Solver is fully deterministic and its source code can be read by a human (without a matrix of floats being seen).

You should take a look at our Food Packaging Quickstart, which seems to map well to your use case.