Weekly Check-Ins and the 1-Meal Rule

Weekly Check-Ins and the 1-Meal Rule

AI coaching is only useful when it improves triage, not when it decorates indecision.

Justin Harris’s coaching note on retatrutide was blunt: 2 mg sharply reduced appetite, added fatigue, and made him more willing to pause or reduce the dose before higher-food phases; the mechanism is appetite suppression. That matters because weekly AI check-ins are not a reporting ritual, they are a triage system: if the check-in cannot separate “hold,” “reduce,” and “change the plan” from one week to the next, it is noise. The falsifiable thesis is simple: the best AI coaching workflow is the one that makes fewer, faster, and more correct weekly decisions than a coach making a vague human check-in.

The actual job of a weekly check-in

Most coaching platforms talk about consistency, accountability, and personalization. Those are outputs, not decision rules. A weekly check-in only earns its keep if it answers three questions in order:

1. Did the athlete’s inputs change enough to matter?
2. Did the body’s response change enough to matter?
3. If yes, what is the smallest useful decision?

That sequence is not fancy, but it is where decision quality lives. The common failure mode is overreacting to data that is real but not actionable, or underreacting to data that is already screaming for a change.

The Joe Webb exchange shows the clean version of this. He reported that on his high day, the same insulin dose as the week prior was dropping blood sugar more than expected, forcing meal 2 to move up by about 30 minutes. He did not report a total derailment, just a change in timing and a need to reduce the dose further on the next high day. Justin’s reply was not “good job being disciplined.” It was a decision: the check-in had identified a real sensitivity shift, and the next action was a smaller insulin dose on the next high day. That is triage: notice the pattern, classify the risk, adjust the smallest variable.

Why AI helps when it triages, and hurts when it summarizes

AI is good at summarizing a week. That is the least important thing a coach needs.

A decent AI layer should be doing something narrower and more useful: mapping incoming check-in data to one of a few decision buckets. In practice, those buckets look like this:

• no change;
• hold and observe;
• reduce exposure;
• shift timing;
• re-order priorities;
• remove the variable entirely.

That is the difference between a chat interface and a coaching system. If the athlete says appetite is crushed, fatigue is up, and the next phase may require more food, the system should not respond with a paragraph of motivational language. It should surface the pressure points and propose the narrowest viable decision.

The Rory Lazowski retatrutide exchange is useful here. Justin said he was trying retatrutide himself because he wanted first-hand experience for clients, but he was uneasy with the idea of forcing appetite lower. He noted it was definitely lowering appetite and that fatigue was higher than normal. Then he said if food was about to go up, he might pause it or reduce the dose until prep; otherwise, run with it and lean out while it is easy. That is not hype around a new tool. It is a triage framework: if appetite suppression helps the current phase, use it; if it starts fighting the phase, back it off.

That logic is exactly what many AI coaches miss. They optimize for “recommendation quality” as if the recommendation exists in a vacuum. It doesn’t. It exists inside a phase, a food load, a fatigue state, and a week-to-week runway.

Weekly check-ins should rank problems, not just list them

A useful check-in does not treat all deviations equally. It ranks them.

For a physique athlete, the difference between two similar-sounding issues can be huge:

• a small appetite change during a stable phase may be tolerable;
• the same appetite change right before food needs to rise may be a real constraint;
• a blood sugar dip that requires moving meals earlier may be a timing issue, not a full program failure;
• the same dip plus worsening fatigue and worsening insulin sensitivity is a larger problem.

This is where AI can improve decision quality if it is built to look for phase mismatch. A weekly check-in is not asking, “What happened?” It is asking, “Did the change happen in the wrong direction for the current phase?”

That is a more coach-like question. It also reduces bad decisions caused by generic rules. A generic rule like “more appetite suppression is always good in a gaining phase because it helps keep body fat down” is too blunt. Justin’s own note on retatrutide was skeptical of the “helpful in gaining” idea. He said he could not yet reconcile that claim and would hold judgment until he had more data. That is the right posture for a coach: treat the current phase as the deciding context, not the drug class, not the trend, and not the internet consensus.

The best check-ins compress uncertainty without pretending to eliminate it

There is a temptation to make AI coaching sound more certain than it is. Don’t.

Decision quality improves when the system reduces the number of unknowns in front of the coach. In the Joe example, the unknown was not “Is insulin sensitivity better?” The unknown was “Does this weekly change require a dose reduction, a meal timing change, or both?” The answer came from the check-in details: same dose, lower blood sugar, meals pulled forward, no day-long overeating. That is enough to make a small decision.

This is the core advantage of structured check-ins. They can convert vague subjective feedback into a tighter action boundary. If the athlete reports:

• appetite is down,
• fatigue is up,
• food is about to increase,
• body comp is leaning out,

then the system should not ask for another week of “data” just because it likes data. It should decide whether the current tool is still aligned with the phase.

In that sense, AI coaching is not about prediction. It is about decision compression.

What coaches should actually want from AI

A coach does not need AI to be inspirational. A coach needs it to make weekly reviews more accurate and less sloppy.

That means three things:

1) It should catch mismatches early

If appetite suppression, fatigue, or blood sugar response changes in a way that affects the next week’s plan, the system should flag it. Not because the numbers are dramatic, but because the plan is sensitive to them.

2) It should keep changes small

Justin’s replies are rarely “overhaul everything.” They are more often “reduce the dose,” “pause for now,” or “run with it while it’s easy.” That’s important. Small decisions are easier to verify on the next check-in, which improves the next round of triage.

3) It should know when not to add complexity

A check-in that turns every issue into a dozen branches is worse than no automation at all. Decision quality goes down when the coach is buried in options. The AI should collapse the week into the few choices that matter.

The practical test for AI check-in triage

If you are evaluating an AI coaching tool, ask a simple question: after a weekly check-in, does it help you choose between hold, reduce, pause, or re-time faster and more accurately than you already could?

If the answer is no, the system is generating commentary, not coaching.

If the answer is yes, you will see it in mundane places first: fewer missed sensitivity shifts, fewer overcorrections, tighter meal timing, cleaner phase decisions, and less arguing with the plan because the plan matched the phase.

That is why the weekly check-in is the real product. Not the interface. Not the chatbot voice. Not the promise that AI “knows” bodybuilding. The product is better triage under uncertainty, and the strongest systems will win because they make the next decision more correct.

Sources Used

• raw/_consumed/2026-05-31/kahunas-export/2026-05-31-w13-18m/transcripts/rory_lazowski___members-c5balaovjbdoeefqmfuqdhh2tbpmfdu16lnf0tnrtmw.md
• raw/_consumed/2026-05-31/kahunas-export/2026-05-31-w13-18m/clients/rory_lazowski___members-c5balaovjbdoeefqmfuqdhh2tbpmfdu16lnf0tnrtmw.json
• raw/_consumed/2026-05-31/kahunas-export/2026-05-31-w13-18m/clients/joe_webb___members-rksigkykimaxwmo_t4_e8nwvbtc2j0etleutkyysads.json
• modules/08-voice/kahunas-coaching-deep-voice.md
• modules/03-knowledge/kahunas-coaching-deep-nutrition.md