T Nation

Sheiko Gold: AI Powerlifting Coaching App

I agree however there should still be variances lifter to lifter. And things should change as the coach knows how you react to certain movements and volume. A crappy coach will just keep you doing the same thing and put 0 thought into your programming.

I put in this program that I wasn’t recovering from what I was doing and it wasn’t adjusted much the next block. Same movements , I think 1 less set weekly on squats, and deads. So 20-1 set for the next block…

1 Like

We should work up a Prileprin app. It would start with optimal volume. From there, you would just keep doing optimal volume. We could get rich.

Lol

I’m currently using Nate’s program app he did and it’s been pretty great so far. I don’t feel like I’m digging a huge hole I have to climb out of week to week.

1 Like

Nate’s write up definitely looked cool, how are you liking it? Anything cool to share?

A few work sets, a bit of assistance work. It’s revolutionary I tell you!

I tried all sorts of high volume stuff and special exercises, it can work but when you are barely recovering you will barely be adapting too. I was already gravitating towards lower volume and I injured my back in June doing some bullshit, that forced me to take it easy on the volume and surprisingly I started making some good progress so I’m sticking with it. A lot of the top lifters don’t do much volume, some people would say it’s all because of drugs and genetics but the IPF guys who train full body 4-6 days a week aren’t all clean either. There is definitely something to it. My deadlift was going nowhere for a while, now I’m hitting PRs on a regular basis (again today).

1 Like

I don’t want to get specific just because I told him I wouldn’t. I have felt fresh and recovered. And I only push 1 lift hard a week so I look forward to that training day to push hard and challenge myself. The rest of the days are spread out for technique and strengthening weak points, building muscle etc. he made a nice program and the spread sheet works very well. There’s a lot more to it than just plugging in your maxes and having a program set up.

1 Like

That’s awesome. I’ve also found myself feeling burned out and low energy outside the gym. Are you working up to a top set and then doing a couple back-off sets?

Pretty much. You can look at my training log if you want. I wouldn’t necessarily recommend that everyone do what I’m doing, the thing is that I’m already big enough and my technique is decent so the focus is not really on hypertrophy. Someone with shaky technique would be better off doing more submaximal work, someone lacking muscle mass would be better off with more assistance work.

You should publish this statement and sell it as an e book

1 Like

Fully customized coaching. $200 a month.

1 Like

Ive had solid results from Blevins AI, joints actually feel better than ever. I ran it for 3 16 week cycles before hopping onto programming by Calgary Barbell. I’m actually interested in the Sheiko stuff but probably will hold out at least for a few months to see if anyone has had some good success with it.

Excessive volume is definitely a problem. But how do you know when it’s too much? Sheiko Gold will know by watching your wellness changes over time in response to the workouts. So it’ll find just the right amount specific to you at that moment in your life. Next year things could be different and it’ll keep pace with the changes you make.

A watchOS version is just about finished. You could leave your phone at home. The classic Sheiko templates have also been added to the Observation Period. So if you’re using those it’s just a few button clicks and you’re ready to go. After 30 days the AI takes over.

“AI in general does not sound like a good prospect for humanity.”

There’s another, less gloomy way to look at AI. The way I see it (I made the app, btw) there’s a lot of “coaches” out there straight ripping people off. So AI can be there to offer you quality at a much more affordable rate. It’ll get you stronger, keep you healthier, and let you keep more of your money in your pocket.

Robert Frederick I presume, nice to have you on here.

That’s a very good question, and everyone has a different story. At one point I bought into Mike Israetel’s MRV concept, and sure I was making progress but I was exhausted after training, progress was slow, and in retrospect I think that while my training was in line with what Israetel advocates I would have made better progress doing less. Now more recently I injured my back and kept volume low on lower body exercises (as per Stuart McGill’s back rehab guidelines) and strangely enough I made some quick progress and I’m actually ahead of where I was pre-injury (nothing incredible, 400 wilks at my last meet which didn’t go well, but at least it’s something).

It seems to me that there is a balance between recovery and adaptation, if all your body’s resources are being used up just to get you back to your previous condition then there isn’t much left to actually adapt and move forward. Of course too little wont work either, so it’s about doing enough to make progress at a reasonable rate and not more than that. And of course that is going to vary from person to person and also between different training styles.

Yeah, my comment there isn’t so much about AI for PL coaching but rather some other issues like replacing workers with machines and such, but that’s another story altogether.

From what I have seen in general, not just with AI, is that there are people out there running fairly large scale experiments with some training methods they came up with or adapted from coaches of other sports. And charging top dollar at the same time. Some of the other AI programs would definitely fall into that category. As for yourself and the Sheiko Gold AI app, I know who you are, I have read some of your articles and posts on the Sheiko forum, it’s safe to say that you know what you are doing. So for anyone interested in using Sheiko’s style of training, this app is likely the best option.

Now the only counterpoint is whether high volume submaximal work is the way to go. As I mentioned previously in this thread, for people with technical issues it is likely the best option as it allows you to practice perfect (or as close as possible) technique over and over rather than pushing close to failure where technique is more likely to break down. But for some lifters, those hard sets that force you to strain are going to be a major factor in making progress. Of course this requires stable technique, at the other end of the spectrum there are programs that have beginner lifters doing AMRAP sets as high as 20 reps…

I used another AI app in the past. Going into I was very optimistic. The program it rolled up for me was based off a questionnaire, I think y’all may have This beat right out the gait having an observation period.

The program had my MRV set extremely high (what I considered high) and was difficult to complete. It had fatigue management options but maxing this out every workout barely lowers the volume or intensity. After sending that block in with maxes out fatigue on every squat and dead lift session it lower my sets per week by 1 on each…I Had to stop using it.

I know this wasn’t Sheiko programming. My main question is how long does it take for the system to regulate a persons fatigue to volume? The other system I could tell going it I would have a rough time recovering however the block (4weeks) was done and no adjustments could be made. This was my main complaint about the program. Now I know I can simply not do as many back off sets in this situation however if I’m paying for a product I don’t expect to have to constantly adjust it when it has my volume basically double what I can recover from.

There are many differences but I’d like to highlight two that I feel are especially important. One is the underlying physiological modeling and the other is the periodization structure.

Modeling: I see MRV as being a useful lexicon to describe certain things that happen to us during training. I find it less useful for making prescriptions for training dose in a particular session.

When I was doing the machine learning on wellness data I recall seeing very low incidence of people doing Sheiko workouts reporting injuries (I’ll dig that up and share official stats in the future). How low? I’m fairly certain it’s under 1%.

That’s not an accident. A central goal with this AI is optimized volume and long term sustainability. An often overlooked route to becoming a champion is simply to outlast your more talented peers. So not getting injured is a huge part of success.

Periodization: I’ve done a side by side comparison of the typical monthly stress in each approach. Both use a mix of low, medium, and high stress workouts, which is good. Assuming the MRV setup is perfectly dialed in with your current state, I see it producing about 25% less high stimulus days per month (on average) compared to the Sheiko AI. Realistically, that could be more or less if the MRV is not dialed in properly (probably the case). The next issue is the distribution of those high stimulus days. In JAI they are concentrated in a cluster while they are more spread out in the Sheiko AI (the exact distribution depends on your readiness). The way I see it, this is both safer and more effective. The statistics I see partially support this. People doing Sheiko vs Other (style unknown) workouts are reporting both greater workloads and improved wellness with Sheiko.

Here’s a brief conceptual overview of Sheiko AI.

Optimized volume is a primary goal. That’s what the observation period helps establish. The app watches what you do each day and how your wellness responds to your workouts. It takes about 30 days to gather enough data to get a good picture of your situation. From there it can start taking over your training. It’ll conservatively try to increase your loading over time, but will keep an eye on your wellness response. If it sees any negative trends developing it’ll cut back on the workload until things improve. Its responsiveness to negative trends is aggressive. So if you want the app to keep giving you harder workouts you’ll need to do your part to stay in good form outside the gym.

Here’s what to expect over a training cycle:

Preparation Period 1: increase muscle mass to raise the potential for strength. Explore variations of the main lifts and search for weak areas

In practice: expect pyramids on the main lifts (at appropriate times in your recovery cycle) and more variations. You’ll also get more accessories for targeted muscle development of lagging areas (all major muscle groups are fair game).

Preparation Period 2: Squash the weakness you discovered in Prep 1 and prepare to lift heavy weights. The goal is the realization of strength potential.

In practice: concentrate on fewer lifts that you’ve found useful. Partial range of motion exercises are particularly useful to help get used to handling near max weights, without the accompanying fatigue of full ROM. More work with the main lifts. Less targeted muscle work.

Competition Period: Initiate deload and taper to competition. Execute the plan to lift heavy on the platform.

In practice: mostly competition exercise work. Few variations. Few accessories. Steadily decreasing volume, same average intensity.

2 Likes

I think we’re actually on the same page here. It only seems like we’re not because people have been running templates that are in Sheiko’s words “approximate programs”. Hopefully people picked the right ones to run and hopefully they knew how to modify them appropriately but I’m going to guess that was not always the case.

High volume for the sake of high volume is not a good approach. What’s a lot better is finding the optimal volume/difficulty for you and the optimal volume/difficulty for me, which are probably different. That’s always been a great ideal to shoot for but really hard to determine in practice. Not so anymore with the AI. It’ll figure it out for us.

1 Like

Yes, that’s true, and I have seen some of his programs like one he wrote for Yury Fedorenko that have plenty of singles with 90%, which is not what people would consider to be the typical Shieko program. Of course the coach himself and general programs written by him are going to be different. My thought is just that some people don’t respond particularly well to low-RPE volume work and would get more out of a few hard sets, does this program make such adjustments?

I remember Sheiko saying that he agrees that the Max Effort method is the most effective way to build strength, but also that he believe it is an extremely high risk for injury, which may explain this:

So yeah, long term gains vs. fast gains that end in injury, it’s not a hard choice to make.

Here’s a question for you:
How does the AI program gauge progress? In all of Sheiko’s programs that I have seen there is a “test day” where you work up to your previous 1rm or attempt a small PR and then the meet is about a month later, how would this work for someone who is only competing once or twice a year?

On a related note, I reckon RTS is the system that would benefit a lot from having an AI solution.

They are already very data driven with TRAC and block reviews plus each of these systems have a big bank of data to draw from.