Nutrition Timing and the Half-Day Rule

Nutrition Timing and the Half-Day Rule

Why most AI coaching gets worse when it changes food too fast, and why the useful move is usually to wait for the next real signal.

Rory Lazowski’s retatrutide note was blunt: 2 mg on Friday wiped out appetite through Saturday and Sunday, and Justin’s response was to hold the dose until he had better data rather than chase every short-term change. The mechanism is simple appetite suppression and downstream food-intake compression. That is the core coaching problem here: if a nutrition change is still unfolding, a second change made too soon can hide the first one. The sharp thesis is that timing matters more than frequency of adjustment, and the best AI coaches should change nutrition on a clock, not on emotion.

The real risk is not under-reacting

Coaches love being responsive. Clients love feeling heard. But the body does not send clean, one-variable reports. Appetite, training performance, GI comfort, scale weight, fluid shifts, and meal timing all move on different timelines. If you change food today because yesterday looked odd, you may end up measuring your reaction to your own intervention instead of the athlete’s response.

Justin’s retatrutide reply is a good example of this restraint. The client reported a dramatic appetite drop at 2 mg, plus more fatigue, and mentioned that if calories were going to rise, he might pause or reduce the dose. Justin did not turn that into an instant escalation ladder. He said he was not sure he reconciled the “helpful in gaining” idea, and then made the practical call: run with it and lean out a bit while it’s easy. That is not hype. It is sequencing. First let the appetite effect settle; then evaluate what the same tool does in a stable phase.

That sequencing matters because appetite suppression is not a neutral variable. If calories suddenly fall, a coach can mistake the resulting weight change, energy change, or meal compliance issue for a different problem entirely. The smart move is to let the first effect declare itself before adding the next lever.

High-day insulin showed the same lesson from the other direction

Joe Webb’s high-day note gives a cleaner nutrition-timing lesson because the signal was immediate. He reported that the same insulin dose as the previous week dipped his blood sugar more noticeably, so he had to bring meal 2 about 30 minutes sooner. He then reduced the dose by 1 IU and still had to adjust timing again on later shots.

The important part is not the drug detail. The coaching lesson is that the change was not “more food” or “less food” in a vacuum. It was meal timing in response to a changing response curve. Justin’s broader nutrition thinking aligns with this: the most useful intervention is often not a dramatic macro rewrite but a small, ordered adjustment that preserves the day’s structure.

That is exactly where AI coaches tend to overreach. A model sees a mismatch and wants to optimize it immediately. In practice, if the athlete is already adapting to one change, the first job is to determine whether the mismatch will persist after the current plan has had time to stabilize.

Food changes should clear a minimum observation window

The clearest published coaching example in the nutrition KB is Justin’s approach to food source tweaks. On the macro front, he is explicit that most results come from nailing macros, and that the visible difference from small source swaps is tiny over short horizons. He says he would be surprised if fruit versus another carb source produced a noticeable difference over a year, although the last few percent matters. More importantly for timing, he gives a practical rule: pre- and post-workout, fruit is fine up to about 50% of the carbs in those meals, and on medium days it is fine in other meals too as long as it is not every meal and at least half the carbs per meal come from more complex sources.

Why does that matter for the daily angle? Because this is a coach who is not making changes meal by meal based on a single day’s appetite or fullness report. He is making a tiered decision: some changes are low-risk enough to allow immediately, while others are constrained by the day type and the broader food structure. That is how you avoid over-adjustment. You define which variables can move now and which variables need time.

The same logic should govern AI coaching interfaces. If the client logs a strange day, the system should ask: is this a true trend, or just a day inside the normal noise band? Did the plan change within the last 24 to 72 hours? Is this a high day, a medium day, a peri-workout meal, or a prep phase where the margin is smaller? Without those checks, the coach will keep “improving” the plan into instability.

Small corrections beat repeated rewrites

One of the most useful details in Joe’s note is that he did not overeat for the day after the insulin adjustment. He simply brought meals closer together. That is the model: preserve the larger plan, change the smallest thing that fixes the immediate issue, and wait for the next check-in before moving again.

That approach is boring, which is why it works.

A lot of bad AI coaching is really just impatience in software form. It sees a client data point and reflexively updates the recommendation layer. But when the body is already in motion, rapid successive changes blur cause and effect. If the athlete is leaning out from an appetite-suppressing tool, don’t also slash carbs the same week because the scale dipped. If the athlete is responding more strongly to the same insulin dose, don’t assume the answer is a wholesale plan rewrite; first adjust timing and dose in the smallest direction that preserves the high day.

What to ask before changing nutrition again

A useful coach, human or AI, should run a short sequence before making another food change:

1. What changed first? Appetite, meal timing, body weight, training output, or digestion?
2. How long has the new input been in place? A day is not the same as a week.
3. Is the current issue caused by the plan or by the response to the plan?
4. Can the fix be smaller? Timing meals 30 minutes sooner is not the same as lowering all calories.
5. Will the next change make the first one impossible to interpret? If yes, wait.

Those questions sound simple because they are. They are also what keep coaches from turning useful data into noise.

The strongest pattern across these examples is not that nutrition needs more precision. It is that it needs better pacing. Justin’s best calls are rarely dramatic. They are usually delayed just enough to get a clean read, then decisive enough to solve the problem without breaking the rest of the plan. That is the standard AI coaching should meet: time the change, keep the signal clean, and stop adjusting before you erase your own evidence.

Sources Used

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