Why does “Next Up” never know, that I already did the wo suggested? It always keeps telling me, I should do the wo “again” that day?
Because it only refreshes every 12 hours
Is it a fixed time or 12h after my ride/wo?
Fixed time I believe, the entire community gets a new rec at the same time I would quess.
Do we then need to “opt out” again? The only “… meaningful improvement” for the majority of users would be having opt out as the default with those who want it needing to opt in
I was on the Build Me Up plan a few months ago, and one of the “Next Up” suggestions was a workout that was on the Build Me UP plan, but it was scheduled and done a couple of weeks earlier.
Not too smart for an AI.
Please put scaling/resolution fixes for the Home Screen into your list. I used to be able to use the entire width of my 21:9 monitor and have smaller icons to see more at once. Even if you ignore the feedback to let us customize what appears where on the screen, this would provide enough real estate that it wouldn’t matter much.
Thanks for the update, but to be completely honest, these personalized recommendations miss the mark entirely. As they stand, they feel inaccurate, incomplete, and too generic to be of any actual use. Basing them on such a shallow pool of information makes the whole feature feel pointless and disruptive rather than helpful.
No reasonable user is going to want to pay for a service that pushes unwanted, poorly optimized suggestions—especially when other platforms on the market already do this so much better. This feels like a massive waste of development resources that could be spent fixing core issues instead of over-engineering a feature nobody asked for. Please give us a permanent opt-out.
Sadly, my guess is the C-Suite at Zwift has been told by their board “AI is everything, you need a metric showing that x percentage of your user base is using your AI features”, so instead of making the platform better for their Customers, they are making it better for Investors (if they go for another round of funding, or go IPO). I fear that once my subscription expires, I will probably go with a permanent opt-out
I really hope one of those improvements will be the option so we can opt out ourself and not have to see it if we dont want to.
I think it’s plain as day that AI features are essential and companies that don’t get on board will be left behind. However that does not require forcing everyone to use those features, especially when the features are immature or don’t serve the needs of everyone. I would have no issue with getting an AI recommendation to do more of the kinds of activities that I enjoy, but that doesn’t include workouts, so there is no workout recommendation that could have any value to me even if it’s executed perfectly.
The other thing about workout recommendations is that they should only be presented to people who are not doing some other kind of training plan, because encouraging people to deviate from the plan is bad. AI recommended workouts are only good for people who have no plan.
They’re at least seen as essential.
Personally I’d much rather see the AI features being improved bots to fill out race fields in quieter time zones. Or better detection of unrealistic performance tied to education about how to improve the setup to more accurate.
Even non customer facing features, such as AI to help unpick the codebase so new features can be rolled out without breaking everything.
They should do all of that as well ![]()
AI in this case is more along the lines of GIGO. If there is no training plan, how can an AI figure out the best workouts to achieve a goal it doesn’t know. If the athlete has a training plan, but Zwift can’t access all the data inputs, would they use Zwift AI for their training (I know I wasn’t when I was training, since it didn’t track my outdoor runs).
If there is a training plan, where Zwift knows the end goal, and has all the inputs (including health data, sleep data, etc), then the AI could be useful:
“I noticed that you skipped yesterday’s workout, and today your sleep hours/score and readiness is still on the low side. According to your training plan, you intended to do a 2 hour interval ride today. I would recommend an hour recovery pace, followed by an hour interval ride instead. This would still let you be ready for your A race in 6 weeks”
This is what I see as the immaturity of the feature. It’s not fatal, but it’s not there yet. It’s OK to release a beta feature for feedback, but it’s silly to have an unfinished beta feature take over the UI. The questions that coaches ask are not optional inputs. It is mandatory to know an athlete’s goals, how much regular time they want to commit, when the events are, the specific demands of the events, the priority of the events, their current fitness level (which may not be in data Zwift possesses). A coach will also give feedback about what kind of improvement might be realistic with the level of commitment proposed.
Back when I first started out as a test engineer, the methodology was “crawl, walk, run” on deployments. That migrated to “Alpha, Beta, Release”, now we seem to be in "Everyone is a beta tester (thanks to Google), where we go from Alpha to MVP (in this case with minimal testing, and no user group feedback) to FulGas out the door.
I don’t pay Zwift to be a beta tester, so having to jump through hoops to have a beta-feature disabled is annoying (at best). Maybe I should invoice Zwift my standard billing rate as a test engineer to test their beta features (I could use the motivation to ride more).
There was an alpha test group and the feedback they gave was that the feature was of poor quality. So of course the next step is to deploy it to every user.
Anyone want to bet on some VP having a bonus tied into getting the feature out the door by a certain date?
It more that, in this case, Zwift try to make us believe they are essential.
That might be true for some. But for me my Zwift experience could very well do without these recemmendations.
The use of AI in the gaming world continues unabated. Zwift is not unique. Devs are convinced this is the path to more money. The gaming world at large views engagement = subscriptions = money. And nothing drives engagement like poorly designed and thought out AI. All those men in those C suites (oh, it’s most likely men) don’t care if you’re mad or happy. They care that you are engaged and giving them money. So, rant away. The money continues to flow and devs will continue to ‘engage’. Ride on.