Look Ahead: See How TrainerRoad AI Helps Today’s Choices Make You Faster Tomorrow

A View of Your Training Future :crystal_ball:

One of the big challenges facing athletes is understanding how today’s training choices will shape fitness in the future. Move or shorten a session, add a big ride, or miss a day and a rigid training plan can feel like one uncertain decision after another hoping for results. That uncertainty creates second-guessing, lost momentum, and less improvement.

That is why we made the TrainerRoad AI Simulation Window.

The Simulation Window gives you a live, 4 week view into your training future while keeping daily choices simple. You can see how changes to your calendar, like swapping a session for a big ride, affects the next 29 days:

  • Watch your future workouts adjust to accommodate the ride.

  • See predicted FTP adjust as you train.

  • Understand why a session is recommended by how it is likely to feel for you.

When you push training harder this week, you also see how the increased training load affects recovery needs in the weeks ahead, so progress stays steady instead of fragile.

Predicted Difficulty

Not knowing how hard a workout will feel or why it was selected for you makes it difficult to understand your training recommendations. It leads to second-guessing, poor preparation, and a higher chance of bailing on your training.

We made Predicted Workout Difficulty to address this.

Predicted Workout Difficulty shows how a workout is likely to feel for you before you start it. With Predicted Workout Difficulty you can better prepare for efforts and understand the recommendation.

Real Life Flexibility

Plans flex with you when life happens. Move a session, add a workout, or drop in a big outdoor ride and see the impact immediately. Your plan reshapes around recovery needs and keeps the next best steps clear.

Meaningful Control

You can see how today’s efforts will affect your future training and make the best decisions to keep you moving forward. Suggested plan adjustments can be reverted at any time, giving you the information, and control to train productively how you want.

Better Preparation

When a workout is expected to be hard, you can prepare: dialing in nutrition, hydration, and mindset. When a workout is expected to be easier, you can know it’s intended to be that way; helping you push harder when it counts and protecting your progress when recovery is needed.

Trust In Your Plan

Understanding why a workout is recommended builds confidence. If your schedule shifts, you can easily pick from the improved alternate workouts menu, to find a workout that fits your time and energy while maintaining your plans trajectory.

What Athletes Are Saying About the Simulation Window

“I found the flexible AI window to be the most valuable. Watching how my workouts changed based on the work I was doing allowed me to understand how each session was meeting me where I was at. Whether I was doing a Threshold session or an Endurance ride, I always found that the AI recommended workout was exactly what I needed on that day and the dynamic changes were spot on.” - Sam L. Beta Tester :woman_technologist:

“Because of the training simulation window, I now feel more comfortable adding off the bike workouts and knowing what impact it’ll have on future TR workouts.” - Chris P. Beta Tester :woman_technologist:

“My training plan included much more ambitious, yet doable workouts. The estimated difficulties were spot-on and really helpful when adjusting the training plan to my circumstances—trying out different options for the next four weeks and seeing how they compare. If I know that a workout is meant to push me close to the limit, I feel less bad when struggling with it—because that training session was meant to be a struggle. The same goes for sessions that were intentionally easier.” - Max L. Beta Tester :woman_technologist:

Together, these stories illustrate the same outcome: less guessing, more confidence, and progress you can feel and measure.

More New Features to Come

This is what training with foresight feels like. You can see the next 4 weeks take shape, understand how today’s choices affect tomorrow’s training, and ride with confidence, knowing every session is chosen to move you forward.

More clarity, more consistency, more Power PRs, and many more features to make you faster. So stay tuned, and sign up at TrainerRoad.com to be ready on launch day!

34 Likes

This sounds amazing. I almost wish I wasn’t going on holiday! But this makes me excited to start training again upon my return in March!

4 Likes

Looking forward to the update - I actually cancelled my subscription due to the lack of some of these improvements. That said, “launching soon” is a bit frustrating. Why not be more specific? If you genuinely don’t know the timeline, why announce it now? Something like “in a couple of weeks,” “next month,” or “Q1” would set customer expectations much better.

12 Likes

I have noticed that TR changes a workout on me if I’ve been for a walk or spent a little longer on the turbo to complete a Zwift route etc. Will the new feature explain why it’s changing the prescribed workout

5 Likes

It’s in beta now. I would imagine that release date is entirely dependent on the rate at which they find and fix release blocker bugs. As to “why announce this now?”, this forum has been filled with salty posts about them not announcing what’s in the new release. Also, they needed volunteers for the beta, so they might as well let us know what’s in the beta so that people who don’t get in aren’t too upset.

11 Likes

As a CPA, whenever I try to use AI for work, I feel like half the time it’s so confident in itself, but so very wrong. The good news, is I’m asking it questions in a field I’m at expert in. With the introduction of AI into the training plans, a field I’m not an expert in, how can I be assured it’s even giving the correct answer?

I guess, I just don’t understand how training plans developed by AI are anything but a downgrade in service levels from training plans designed by humans. AI today, is literal slop. It acts like it’s right, but usually is not.

8 Likes

AI is only as good as the data it consumes. Gen AI, which I assume you’re talking about, takes a ton of data to spit out GENeral information. Unfortunately sometimes that info it consumes isn’t correct so it gives you those incorrect answers. I’ve had better luck in telling it to double check or use multiple sources.

This AI they’re talking about is only fed training data so should be more reliable there. Clearly I don’t know exactly what they’re doing but my guess is the closer you are to the “average” cyclist the more accurate it’ll be.

More on topic here I hope we’re not getting an announcement post a day on this. Frustrating when this kind of thing is teased before it’s available or at least doesn’t have a date with it. Hoping I get thrown in the beta group sooner rather than later though. Recently came back to TR after a few years off and now that I know a bit more about training I’m often questioning the adaptive training.

1 Like

The platform already uses “AI” (often under the more appropriate term “machine learning”), eg for FTP detection (and presumably also for train now, and plan builder). The new “AI” branding is just popular right now, and is used to distinguish the new features and ML models that they’ve announced.

1 Like

Will the new features allow ‘conversational’ interaction with the AI? E.g. could I ask it to review some recent threshold workouts with me, or if there were any signs in my data ahead of a recent over-training injury?

This sounds AWESOME!!!

Joe

1 Like

Will this work with any workouts on your calendar that aren’t tied to a plan generated by TR (I.e. Our own custom made plans?). That will be really useful if so. If it’s tied to a plan generated by TR, like most features are now, less so.

I have no inside information but I would be surprised –at this point– if there is conversational interaction. I do think that is probably on their radar. But who knows …

Has Trainer road managed to find a way to analyse outdoor and other imported sessions ? I’m just wondering how this will account for past and historical data when not from the trainer road platform. I potentially thinking of coming out of 1 to 1 coaching and back but in the past the platform could not integrate many sessions well and would give them crazy grades which would throw off the whole premise of the platform.

2 Likes

Per the Instagram comments there will not be any conversational part to the “AI”. It’s not an LLM, that’s just one very specific type of “AI” (really not the proper term, but when a term gets misused enough it takes on the new meaning, see “literally” ).

4 Likes

Ah, thanks. It’s be a cool feature eventually. Like a real ‘ai coach’

OK. This actually sounds great, and something along the lines I was expecting when I returned to TR a year ago after being on other training platforms. I ride outside whenever reasonable and don’t do structured workouts when I do. I want a forward looking plan that fully takes these outdoor rides into account and shows me what I should be doing in the future in light of what I’ve done, including all indoor structured workouts and all unstructured outdoor rides. The adaptive plans, RL/GL, Solo Rides, Group Rides… seemed like a hodgepodge of disjoint features. I quit TR as a result planning to try something different this winter. Now, I may well come back.

2 Likes

TR has been able to incorporate outside rides to impact AiFTP, RL/GL for a while. It does not (prior to the current beta version they are talking about) impact progression levels. So it isn’t perfect, but it also isn’t true the the current platform can’t deal with outside rides. Hopefully this new launch will make it even better

2 Likes

The proper term (way out of vogue) for what TR AI / LLMs are is “M/L” (machine learning). None of these systems are “AI” (artificial intelligence). But M/L went out of vogue a decade or more ago.

2 Likes

I just listened to the low cadence podcast this morning and was happy to hear Jonathan say AI and then say something like, “really, it’s Machine Learning”.

4 Likes