Often I hear people confuse “predictive” with “preventive.” Until one day I found this image, since then the people I show it to don’t confuse it anymore.
Reactive: Fix it when it breaks
The problem has been identified, it is now a matter of limiting the damage and especially the costs caused by it. In the case of project management, my project is late and I have to minimize my losses by preparing good claims or counterclaims.
Preventive: maintain it at regular intervals so it doesn’t break. In the context of project management, the delay has not yet crystallized (the schedule is still up to date and the delay is quantified), it is about reducing the risk of project failure by regularly eliminating large frictions on the project in order to release stress in the project.
Predictive: predict exactly when it will break and maintain it accordingly. In the context of project management, this means predicting the risk of drift on a risk-by-risk basis; even suggesting an approach to avoid the risk. For this, a “clean” historical base (digestible by the machine) is necessary in order to train the machine.
Why this differenciation is critical?
Often when the three approaches are confused, launching an artificial intelligence approach seems to require immense means to create this “clean” database: change of habits, change of tools, change of processes. However, by focusing on the other two niches preceding this major transformation, it is possible to consider a much less major deployment while reaping a significant ROI.
At Lili.ai, we have been working for 4.5 years to minimize the friction points of a deployment while focusing on showing immediate value to the user. Discover how Lili.ai can make you save money in a few days.