New trainer road athlete here. I don’t test well, so I was thrilled to hear about the AI FTP detection, but I just realized that only indoor workouts count…and I almost exclusively ride outside.
Any updates on whether the software will learn from outdoor rides anytime soon?
No real answer, but the outdoor rides with a powermeter will count the AI software will learn from them after you do 10 indoor rides. My theory is that the system at the moment needs that initial consistent base (10 indoor workouts) before it can carry out minor tweaks with either outdoors or indoors data.
Latest I’ve heard was on the podcast a month or more ago, that they were hoping to get into a private beta (small test group of people outside of TR). Nothing new since then, so I wouldn’t get my hopes up with anything coming ‘soon’.
They need 10 indoor TR workouts. AKAIK, it doesn’t matter which ones. You could knock out four 30 minute sessions in 2 hours if you wanted. Do that twice and you are almost done. Once you have done that, outside workouts count.
Hey @K_Alford! Welcome to TrainerRoad and the TrainerRoad forum!
I’m glad you are excited about using AI FTP detection 
.
AI FTP Detection will work for you if you only do unstructured rides with power data. However, at the moment, we need you to complete 10 indoor TR workouts for initial calibration. After those rides, all of your rides can be outside, and it’ll still work.
While I can’t say when it will be ready, the team are currently working on a version of FTP detection that’ll remove the 10 TrainerRoad Workout requirement while maintaining accurate detections.
Let me know if you have any questions for me about this or anything else while you are getting started with TrainerRoad! My DM’s are always open too 
.
Does AI FTP Detection, once it gets those initial 10, then look back at your whole training history? I did as was suggested above, and did 8 15 minute workouts, and my FTP was nailed to exactly what I would have expected. I’d done 2 other short TR workouts the day before as well.
Yup as @mcneese.chad says, AI FTP Detection is looking at your entire training history - inside and outside rides!
Thanks Sarah. Will there be a difference in the accuracy if you mostly (or only) feed it outside rides?
This change surprised me, since the beta version of AI FTP worked great with me almost exclusively doing outdoor rides (I often incorporate structure into my outside workouts). What changed about the process when it was released?
We only need you to complete 10 indoor TR workouts for the initial calibration. After those rides, all of your rides can be outside, and it’ll still be equally as accurate!
That’s great! This last update improved the accuracy of AI FTP Detection for specific use-cases and we’re working on an iterative release process to expand use-cases. There are some technical details for the outside-only detections that are in testing and will be in production soon.
Feel free to delete my last post. I’m just frustrated that one kind of workout that would be super easy outside is ridiculously hard/a different kind of workout inside, and that there’s no difference in indoor and outdoor ftp. But I don’t need to participate in the forum—I was just venting and I won’t do it again. I’ll be patient for updates.
Thank you!