Insulin Timing and the 30-Minute Rule in High Days

Why small nutrition changes need a waiting period before you change them again

The clearest coaching signal in the set is simple: when Joe Webb’s usual insulin dose started dipping blood sugar on a high day, the fix was to reduce the dose by 1iu and bring the next meal 30 minutes sooner, not to keep stacking adjustments on the same day. That is a timing problem, not a heroics problem. The mechanism is basic absorption mismatch: if insulin action is now outrunning incoming food, you change the schedule or the dose once, then let the new pattern show you what actually changed. The falsifiable thesis here is that most nutrition adjustments fail not because they are wrong, but because they are made again before the first change has had time to reveal its effect.

In the Joe case, the signal was specific and repeatable. On his high day, the same insulin dose that worked the previous week now pushed blood sugar down enough that meal 2 had to move up by 30 minutes. He tried a 1iu reduction for meal 3 and felt fine, then hit the same issue again on the next dose and had to pull the following meal forward again. He did not overeat to compensate. That matters because it separates the problem from appetite, compliance, and total daily intake. The immediate lesson is not "change everything." It is "change one lever, observe the next repeat, then decide whether the pattern is fixed or not."

That is the part most AI coaching systems get wrong when they are optimized for responsiveness instead of restraint. If the model is trained to answer every message with a new tweak, it will create fake precision. In real coaching, the useful move is often to stop at the first adjustment and wait for the next data point. Nutrition timing changes are especially vulnerable to overreaction because the body may already be in motion from the earlier change, but the coaching dashboard makes it feel like nothing has happened yet. The result is not faster progress. It is a pileup of tiny corrections that you can’t separate later.

Justin’s nutrition guidance in the KB keeps showing the same hierarchy. First, the big variables: macros, total intake, and the actual periodization structure. In the fruit example, he says he would be surprised if someone saw a noticeable difference over a year even if all carbs came from fruit, because most of the result comes from nailing macros. The details still matter, but as the last few percent. Pre- and post-workout, fruit is fine up to about 50% of the carbs in those meals, with no downside. On medium days, fruit 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. On high days, carbs are so high and insulin elevated all day that sugar barely matters and fruit helps keep food volume down.

That framework gives you a clean rule for timing adjustments: make the smallest change that matches the day type, and only after the day type itself is stable. If the high day is the relevant context, change the high day, not the entire week. If the issue is a meal-level glucose dip, move the meal or reduce the dose tied to that meal. If the issue is only that the plan has not yet had time to show its effect, do not stack more corrections on top of it. Coaches love calling that "being proactive." Usually it is just impatience.

Justin’s body-comp message in the Rory exchange points to the same restraint from the other direction. When retatrutide knocked down appetite hard, he said he did not love the idea of forcing appetite lower, but he would run with it and lean out a bit while it was easy. That is not the same as changing the dose every time fatigue or food intake feels off. It is choosing a phase-appropriate lever and letting it work. If calories are going up, he was open to pausing or reducing the dose. If the phase is still a lean-out, he preferred using the appetite effect to your advantage. The key is that the decision was anchored to the phase, not the last text message.

This is where AI can actually be useful: not by inventing better judgment, but by protecting the interval between judgment calls. A good assistant should ask, "What changed first?" and then shut up long enough for the answer to matter. Did the same insulin dose start dipping glucose after a high day? Did appetite suppression arrive after the first retatrutide dose? Did fruit placement change meal tolerance? Those are timing questions. They deserve a delay before the next adjustment.

For coaches, the practical policy is straightforward:

Define the lever in the smallest possible unit. Dose, meal timing, carb source, or day type.
Make one adjustment that directly matches the signal.
Hold long enough to see the next repeat of the same condition.
Only then decide whether the first change worked, under-shot, or overshot.

That sounds almost too obvious, but the field is full of people who violate it every week. They make a correction on Monday, then a different correction on Wednesday because the Monday correction has not yet produced a dramatic visible change. In nutrition coaching, that usually means you end up measuring your own impatience instead of the client’s response.

The best evidence in this KB does not support reactive churn. It supports phased restraint. Macros beat food purity over long horizons. Day type beats one-off food dogma. Meal timing fixes a meal-level mismatch better than a whole-plan overhaul. And in the only hard example with a number attached, a 1iu reduction plus a 30-minute meal shift addressed the immediate glucose issue better than continuing the prior setup unchanged.

If you want an AI assistant to help with nutrition coaching, ask it to do less, not more. It should surface the relevant variable, remind you what phase the client is in, and resist the urge to optimize after every message. The coaching edge is not constant adjustment. It is disciplined delay. That is how you avoid turning a normal adaptation into a self-inflicted problem.

Sources Used

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