Remix.run Logo
nerevarthelame 4 days ago

I ran into a handful of technical issues, which are understandable at this early stage. But on a more fundamental level, the exercise recommendation doesn't seem like it demonstrates a good knowledge of fitness program design. So I really would not recommend anyone use it to design a fitness program, at least in the current state.

It might be better to focus not on generating routines from scratch for users, but rather logging existing workouts, and importing routines/templates other people have designed. (I know you expressed interested in working on the latter in the comments here.)

Some specific issues I encountered:

1. I said that I wanted a full-ish body workout, and it recommended 33 distinct exercises for a single session. That's wildly impractical.

2. The exercise selection seems to just pick 3 random lifts (based on available equipment) for each selected muscle group and ignores the fact that many lifts hit more than one muscle group at a time. It also disregards the widely accepted notion that certain muscle groups respond better to different training volumes.

3. The selection of specific exercises seems arbitrary, recommending some that are very far out of the mainstream and with poor resistance curves.

4. It recommended exercises with equipment I do not have (I'm a home gym person - no way to exclude machines)

5. It suggests strange branded equipment (e.g., "Dynaband Shoulder Press")

6. If you go back to select different equipment, it will continue to recommend exercises with that equipment.

7. Can't delete recommended exercises, nor even add new ones.

beezle 4 days ago | parent | next [-]

Similar take. The UI and lift explanations are nice but the lift selections (when even available) are a bit dubious at best. In some cases, sub-optimal, in other cases out of left field.

My personal suggestion would be to populate the data base with all the standard primary/secondary lifts (bar, dumbell, machines) and let users build their own program. Maybe have an option to suggest alternatives for any given lift.

After that is done, then maybe work out routine creation, hopefully with some general input to the developer from actual trainers.

Also agree, keep the equipment simple/standard and avoid branded machines,etc.

surgomat 4 days ago | parent [-]

Thanks for the honest feedback.

Yeap ! Once that foundations are stable: introduce routine creation tools, ideally shaped with input from "real" trainers and a free "marketplace" to pick any workout session / program, provided by the community.

I guess we are aligned, lol :) Thanks again !

surgomat 4 days ago | parent | prev [-]

Wow, Thanks a lot for taking the time to write this seriously. This kind of detailed feedback is incredibly valuable, especially at this early stage. Thanks !

So, yeah you're absolutely right: the exercise suggestion logic is very basic for now, and doesn't reflect proper programming principles (volume, movement patterns, recovery, compound/isolation balance, etc.). It’s more of a discovery tool than a real "smart" coach let's say. I probably need to make that clearer in the current UI.

To your specific points:

1. 33 exercises for a single session YES that's overkill, lol.

2. It currently just picks 3 per muscle group, with no upper cap or "contextual logic" let's say, and that'll change soon.

3. I'm already working to "categorize" the exercises by introducing metadata like compound/isolation, primary/secondary muscle groups, movement patterns, and tags for resistance "quality" let's say or "popularity".

4. I’m not sure I fully understood your expectation here. Do you mean you'd like to explicitly exclude certain machines or "types" of equipment, even if others are selected? (e.g. : "I have dumbbells and a pull-up bar, but please exclude all cable or machine-based exercises"). If so, that makes total sense. I'd love to clarify and improve that part of the UI. Thanks.

5. OK

6. Changing equipment doesn’t fully update the list : bug confirmed, I'll fix that. (For the moment yes you have to do it 2 time - render problem).

7. It's high on roadmap. You'll be able to fully edit your routine very soon.

Really appreciate the critique. No ego here I want this to be useful and respectful of good training principles. If you're ever up for helping shape that direction (even just high-level ideas), I'd love your input, what do you say?