Four filters to ask yourself about the next AI project you want to start:
I've developed four filters to evaluate which AI project to start with: it should be small, painful, repeating, and have readily available data. Most projects fail the fourth filter, but when yours passes all four, you've found your starting point.
- Is it small?
- Is it painful?
- Does it repeat?
- Do you already have the data?
A client walked through these and picked purchasing. Their reasoning: buy price and sell price are what control margin, so start where the money already moves. The first build is a cost book out of the supplier price book they already have.
Most projects fail the fourth filter. The work is small and painful and repeats, but you can’t get the data easily. Many of my clients run Datacor and without their Data Lake, it can be a challenge getting the data direct. However, there are ways around automating the data extract.
Run your own list through the four. When your project passes all four, that is where you start now (and message me).