Sad to hear this to-be feature gets canned.
This would be awesome if this would work out, but it seems AI is not the best solution to do implement this.
To be honest i really liked the TrainerRoad featute that suggested the next workout based on the latest ones. It was very accurate back when I tested it two years ago.
Maybe there will be another kind of middleware to implement the suggested feature
Even though TR calls it AI, on a Youtube video, Nate explained they used machine learning to analyze the data. From that analysis I am sure they just created rules, however sophisticated or not. People seem to like this “AI” (rules) so I am assuming it’s reasonably well done, but I would not be surprised if it still did not maintain a lot of the TR mindset around training (TSS / Sweet spot focus). The big difference being taking feedback and modifying based on feedback (which is great).
Our thinking is so different here than TR’s that we need to find different ideas around this. Our TODAY feature has a bit of intelligence but is in no way a substitute for a plan or like their AI researched rules.
Coach Andrea has spent a lot of personal time trying to get different AI tools to produce a good plan but has so far failed, the same as I did.
We will continue to invest in TODAY and improve the algorithm I am just waiting for that incredible idea to come from someone or some where on how to do it 
I’m excited about this feature and more than willing to be a beta tester! This training method will be perfect for someone like me who travels frequently for work. Count me in!