ChatGPT "coaching"

[gauge22-v2]

Upgraded to paid CHATGPT subscription ($20/month). Bolded lines are the instructions I gave to the LLM. I tried to document these as I went, but likely missed a few.


CHATGPT What is my bodyfat as per calipers and pictures. Compare and average. I put in my stats and uploaded recent pictures and 4 points of body fat caliper measurements. Caliper measurements were the ones I could reach. Thigh, ab, iliac crest, chest.

CHATGPT Calculate my TDEE based on this information.

CHATGPT Attached is my nutrition log since May 15th with total calories, protein, carbs, and fat. Recalculate TDEE and compare to last query.

CHATGPT Attached is my total sleep time. I used average 7 day total time from Garmin to July 28th. From there forward I am using OSCAR

CHATGPT create tracking spreadsheet for capturing this information that I can upload on a weekly basis containing columns for bodyweight, calories, protein, carbs, fat, steps, sleep. Also include calculated 7 day averages for TDEE and caloric surplus/deficit. Call this information “Daily Log” and display as a tab in Excel.

Tab above took 5 revisions. I compared to my own tracking sheet for weight, TDEE etc. I also had the columns reordered and made some minor adjustments

CHATGPT Create additional tab for tracking bodyfat with 4 point caliper readings. Call this tab “Body fat tracking”

CHATGPT Create additional tab for workout tracking call this “Lifts Log”.

This tab also took a few revisions to make it more user friendly and easy to populate from the data I capture in the gym using Fitnotes.

CHATGPT Create a lean bulking plan. My goal is to limit fat gain to ensure I am no more than 10% bodyfat at the end of 12 weeks. Provide macros for food by week

Updated starting body comp (at 208 lb, 8.5% BF)​

• Fat mass: 208 × 0.085 = 17.68 lb
• Lean body mass (LBM): 208 − 17.68 = 190.32 lb

Forecast to stay ≤10% by Week 12​

Assume a slow gain (~0.5 lb/week → ~+6 lb in 12 weeks) and typical lean-bulk partitioning.

• If ~35% of gain is fat:
  • Fat gained ≈ 2.1 lb → FM ≈ 19.78 lb → End weight ≈ 214 lb → ~9.2% BF
• If ~40% of gain is fat:
  • Fat gained ≈ 2.4 lb → FM ≈ 20.08 lb → ~9.4% BF
• If ~50% of gain is fat (conservative upper bound):
  • Fat gained ≈ 3.0 lb → FM ≈ 20.68 lb → ~9.7% BF

Conclusion: with the current slow-gain plan, you’re comfortably under 10% at 12 weeks. Even an 8 lb total gain (216 lb) at a high 50% fat share lands ~10.0%—so keep weekly gain ~0.5–0.6 lb to stay safe.

Weekly calories & macros (keep these unless your 7-day trend says otherwise)​

Protein fixed at 1.2 g/lb ≈ 250 g/day. Fat ~80 g/day. Carbs scale to hit calories.

Week	Calories	Protein (g)	Fat (g)	Carbs (g)
1	3,050	250	80	332
2	3,100	250	80	345
3	3,150	250	80	358
4	3,200	250	80	370
5	3,200	250	80	370
6	3,250	250	80	382
7	3,300	250	80	395
8	3,350	250	80	408
9	3,350	250	80	408
10	3,400	250	80	420
11	3,450	250	80	432
12	3,500	250	80	445

Adjustments:

• If >0.8 lb/week → −100 kcal (mostly −25 g carbs).
• If <0.3 lb/week → +100 kcal (+25 g carbs).
• If abdomen skinfold rises >2 mm over 2 weeks → −150 kcal next week.
• Poor sleep week (<6.5 h avg) → hold calories steady.

Quick reminders​

• Keep StepMill ~20 min/day, and your current DC/Myo split.
• Log in the v9 tracker; I’ll keep recalculating 7-day TDEE and surplus/deficit to keep you on track for ≤10%.

XLS Tracking headers:


Daily Log



Formulas are in columns K and L

Weekly Summary

Formulas are in columns B-H and N-Q

Conclusion:

Very user friendly however you have to think about the outputs critically as I caught a few major mistakes. IE: Subbing in fat for protein at one point...also no chance I am at 5-6% bodyfat, so I juiced that upwards. I will start this lean bulk September 1st and provide the weekly updates to the LLM. The updates will include pictures and measurements. I am considering also adding in tape measurements for key bodyparts (thighs, waist, etc etc).

 

[KingOfSushi]

Yep. It’s a phenomenal resource. Not much more to say since it’s mostly already been discussed in the other thread covering this topic.

 

[UnholyAlliance]

ChatGPT may well be taking over the world, but as you now know it's not quite there yet but very close!

A great resource in the right hands and there are offline LLM's out there (the LLM database is quite large!) that you can set up on your computer if you care about privacy, especially if you want to experiment with AI.

 

[Dutch1909]

if can be a very helpful tool, but it depends how you use it. A younger guy at my gym got this advice for his new cycle through ChatGPT:
1000mg test e
500mg tren a
500mg EQ
500mg mast e

It’s his 2nd cycle

 

[AllOkJumpmaster]

I will say this, I coach on the side, and I am very glad it is not my only source of income the way things are trending. That said, I think, at least for the time being, some people will always want a human coach that has experience, at least until another generation completely flushes out, and human race is one that has completely grown up with AI from Day. Though here are some things I do like about this, is that I only have so much time to give, so that keeps my client load within a range to ensure I can give them the attention they deserve. Some people, how do I say, pay me far too much for what they need. I give them the attention I believe they deserve, but some people that are not in comp prep, just require such little management, and adjustments. They just dont need a coach. These types of things can provide what they need and they can save a lot of money.

 

[Siggy]

if can be a very helpful tool, but it depends how you use it. A younger guy at my gym got this advice for his new cycle through ChatGPT:
1000mg test e
500mg tren a
500mg EQ
500mg mast e

It’s his 2nd cycle

Chat GPT out here trying to kill people. I wonder what it would suggest for his third or fourth cycle.

 

[PissedPajamas]

It’s been pretty good at guiding my offseason and not just throwing calories at me when I’m hungry. Settled into a good .5-.8lb rate of gain weekly after factoring in the bloat lord saturation arc (gained like 10lbs in a week from adding in Deca, anadrol and EQ after raising test to 1g lol).

It’s been very unbiased in my weekly check-ins, I enjoy it for the tool it is

 

[luki7788]

Honestly, if someone uses a method where ChatGPT tells you to increase calories every week and promises muscle gain and minimal fat gain - that’s already a setup for failure. It just doesn’t work like that. We’ve been doing this long enough to know weekly calorie increases don’t make sense.

As for coaching Chat can be fine for general lifestyle advice. If you ask smart questions and give it the right info, it might give decent suggestions. But it’s never the same as working with a real coach.

And if you’re prepping for a show? No way it can guide you properly. It’s way too individual. You need years of experience, instinct, and that “coach feel” which no textbook knowledge or AI can replace. Real prep coaches aren’t going anywhere because of ChatGPT.

 

[johnjuanb1]

Honestly, if someone uses a method where ChatGPT tells you to increase calories every week and promises muscle gain and minimal fat gain - that’s already a setup for failure. It just doesn’t work like that. We’ve been doing this long enough to know weekly calorie increases don’t make sense.

As for coaching Chat can be fine for general lifestyle advice. If you ask smart questions and give it the right info, it might give decent suggestions. But it’s never the same as working with a real coach.

And if you’re prepping for a show? No way it can guide you properly. It’s way too individual. You need years of experience, instinct, and that “coach feel” which no textbook knowledge or AI can replace. Real prep coaches aren’t going anywhere because of ChatGPT.

I’d trust @luki7788 to be my coach over ChatGPT! Best way to choose a coach, in my humble opinion, is to look at their physique! If they can’t do it to their own body, then they aren’t going to be able to do it to my body.
Walking the walk = RESPECT!
Anyone can talk the talk out their arse!
A.I. is a great tool that should be viewed as an additional resource.

 

[bbxtreme]

I will say this, I coach on the side, and I am very glad it is not my only source of income the way things are trending. That said, I think, at least for the time being, some people will always want a human coach that has experience, at least until another generation completely flushes out, and human race is one that has completely grown up with AI from Day. Though here are some things I do like about this, is that I only have so much time to give, so that keeps my client load within a range to ensure I can give them the attention they deserve. Some people, how do I say, pay me far too much for what they need. I give them the attention I believe they deserve, but some people that are not in comp prep, just require such little management, and adjustments. They just dont need a coach. These types of things can provide what they need and they can save a lot of money.

ChatGPT is very good at learning you and telling you what you want to hear. That is the exact opposite of what competitors need in a coach as they tell you what you don’t want to hear.

The human element is not going anywhere IMO.

 

[AllOkJumpmaster]

ChatGPT is very good at learning you and telling you what you want to hear. That is the exact opposite of what competitors need in a coach as they tell you what you don’t want to hear.

The human element is not going anywhere IMO.

Yeah I was thinking more about it, and I agree more with this. There is also 2 other components with the human element at play.

1. People paying upwards of $250+ (sometimes A LOT more) a month for coaching are typically much more likely to remain invested to the plan and process i.e. if I am paying this much, I better do the shit, rather if I am paying $20 a month, ehh thats a Netflix subscription, If I dont take it as seriously, its whatever.

2. Would be the fact that coaches learn their peoples tendencies and behaviors over time. Similarly to what you said, but you can tell when a guy needs to pull back, or suck it up and push forward. You can tell when someone is not following the plan and not being truthful, or maybe doing more than you are telling them. Learning when someone is not following the plan is more than just the scale not moving in either direction, you learn through communication with them, and you can just pick up on things. I feel like ChatGPT might just asses and just be like oh the scale isn't moving, reduce calories etc.

That said, I am stunned people are using for drug protocols though. Can you submit bloodwork to it to have it asses your health before recommending compounds / dosages?

 

[P30LEM]

I was putting some stuff in there the other day and it recommended a diet that was 300g protein, 275 carb, and 80g fat, and told me it was only 2090 calories

 

[TheOtherOne55]

ChatGPT is very good at learning you and telling you what you want to hear. That is the exact opposite of what competitors need in a coach as they tell you what you don’t want to hear.

The human element is not going anywhere IMO.

This 100%

As amazing as the plan ChatGBT gave gauge22-v2 is, that is still quite a bit of effort for someone, competitor or not. The amount of upfront info a user has to put in is just a lotttt. Compare that to the dead simple plan coaches like Justin Compton give where each meal is just: 8oz chicken, 200g rice, 100g green veggies. 30 mins cardio. Just shows you that this specific fix works well for a very specific type of client...if you are willing to put in the effort AND somewhat know your shit already. The new client, new to fitness, trying to figure it out cant handle this, at least in my opinion.

 

[Sectør]

They just came out with ChatGPT-5 too which is incredibly powerful, exponentially more than previous versions.

I don’t use it to tell me what to eat or how much, I mainly just keep the same meals everyday then make adjustments based off a combination of the mirror or scale.

I do find it very helpful for learning about supplements, or medical questions, or even some of these new mitochondria peptides—dosage, clinical trials, the pros and cons, synergistic stacks and interactions, the the pharmacokinetics of how things work.

But it’s great for any topic of interest. You can start with a general question about anything then do a deep dive into basically any topic. It can also be used to pull massive amounts of data then refine it with set filters or parameters. I’ve found some great stocks and investments through it.

Can also be used to help automate a lot of processes or tasks for work. I’ve had it write VBA editor code for me in excel to automate a lot of spreadsheets. Can only imagine what it’ll be able to do 5 years from now

 

[P30LEM]

ChatGPT is very good at learning you and telling you what you want to hear. That is the exact opposite of what competitors need in a coach as they tell you what you don’t want to hear.

The human element is not going anywhere IMO.

I agree. AI is still extremely limited despite how useful it is and everyone having a hard on for it. If you know what you're doing you can get it to spit out literally any answer you want just by how you phrase the prompt or question.

 

[gauge22-v2]

I think the value is going to be the tracking and almost like "business intelligence" but for your diet/workouts. Now that the excel sheets are setup (I can dump out of cronometer and fitnotes with cut/paste) it will be simple to populate...

I had it generate forecasted bodyweight etc based on calories and create a quite amazing chart with linear regression, smoothing etc....

I think it is super intriguing...but given my job is a CFO and I live in data and spreadsheets........it has a certain appeal to me beyond the lifting aspect. Yes, I proudly acknowledge I am a data nerd....

 

[gauge22-v2]

They just came out with ChatGPT-5 too which is incredibly powerful, exponentially more than previous versions.

I don’t use it to tell me what to eat or how much, I mainly just keep the same meals everyday then make adjustments based off a combination of the mirror or scale.

I do find it very helpful for learning about supplements, or medical questions, or even some of these new mitochondria peptides—dosage, clinical trials, the pros and cons, synergistic stacks and interactions, the the pharmacokinetics of how things work.

But it’s great for any topic of interest. You can start with a general question about anything then do a deep dive into basically any topic. It can also be used to pull massive amounts of data then refine it with set filters or parameters. I’ve found some great stocks and investments through it.

Can also be used to help automate a lot of processes or tasks for work. I’ve had it write VBA editor code for me in excel to automate a lot of spreadsheets. Can only imagine what it’ll be able to do 5 years from now

Absolutely! ChatGPT 5 goes much deeper.

One of the not workout things I used it for today was to suggest a good hike within a 45 minute drive of me, elevation requirements, distance, with a ridge walk...it gave me 4 really good options. I then asked which would be "dog friendly", ie limited rocks/scree, and it dropped 2 of the 4.

Endless possibilities.

 

[Sectør]

I think the value is going to be the tracking and almost like "business intelligence" but for your diet/workouts. Now that the excel sheets are setup (I can dump out of cronometer and fitnotes with cut/paste) it will be simple to populate...

I had it generate forecasted bodyweight etc based on calories and create a quite amazing chart with linear regression, smoothing etc....

I think it is super intriguing...but given my job is a CFO and I live in data and spreadsheets........it has a certain appeal to me beyond the lifting aspect. Yes, I proudly acknowledge I am a data nerd....

Ayee we work in the same field, good stuff. Although I’m not CFO level yet

 

[alfresco]

The Singularity Is Near by Ray Kurzweil

 

[KingOfSushi]

Honestly, if someone uses a method where ChatGPT tells you to increase calories every week and promises muscle gain and minimal fat gain - that’s already a setup for failure. It just doesn’t work like that. We’ve been doing this long enough to know weekly calorie increases don’t make sense.

As for coaching Chat can be fine for general lifestyle advice. If you ask smart questions and give it the right info, it might give decent suggestions. But it’s never the same as working with a real coach.

And if you’re prepping for a show? No way it can guide you properly. It’s way too individual. You need years of experience, instinct, and that “coach feel” which no textbook knowledge or AI can replace. Real prep coaches aren’t going anywhere because of ChatGPT.

I just told it the rate I wanted. The key in having success with it I’ve found is giving it as much data as possible and bouncing the knowledge you already have off of it. When I originally told it my goals for example it recommended adding 150-200 calories if my weight stalls for 2 weeks.

Its been a great tool for my self coaching, but the user needs to have the base knowledge themselves imo.