Weekly Check-In Triage: 5 Signals from Justin Harris

Weekly Check-In Triage: 5 Signals from Justin Harris

Decision quality in AI coaching comes from sorting noise fast, not sounding smart.

The strongest practical signal in the Kahunas coaching corpus is simple: when insulin sensitivity shifted on a high day, Justin Harris told the client to reduce the insulin dose further rather than force the day to fit the old plan. That is a triage mechanism, not a philosophical one: compare the current data to the last decision, then change the smallest variable that restores execution. The falsifiable thesis is that weekly check-ins get better when AI coaches are used as triage engines for decision quality, not as prediction machines, and worse when they turn a clean signal into a longer debate.

Most weekly check-ins are not missing information. They are missing ranking.

A client reports better insulin sensitivity, the same dose now dips blood sugar more, meals need to come closer together, and the obvious move is to lower the dose on the next high day. That sequence matters because it shows the core job of the check-in: identify the variable that changed, identify the constraint it created, and make the adjustment before the next cycle. The coach did not wait for a perfect trend line. He did not ask for a six-paragraph theory. He used a live response to one repeated stimulus and updated the plan. That is how good triage works: one sign, one constraint, one decision.

This is where AI coaching can either help or clutter the process. The value is not in producing a polished explanation for every note. It is in noticing which updates actually change the next week. If a client says the same insulin dose is now too much for the same meal pattern, that is not “interesting context.” It is an actionable boundary. If another client says retatrutide at 2 mg made appetite disappear and fatigue rise, that is also not trivia. Justin’s response was not to worship the drug or reject it on principle. He said he was trying it himself, did not like the idea of forcing appetite lower, and would hold off on a verdict about “helpful in gaining” until there was more data. Then he leaned toward using it while the athlete was in a phase where leaning out was the immediate priority. The mechanism here is appetite suppression plus fatigue, and the decision quality lesson is sharper than the drug discussion: use the check-in to determine whether the current tool matches the current phase.

That same logic shows up in the off-season nutrition discussion. Justin’s stated goal is not simply “eat more.” It is to teach the body to digest and assimilate a massive amount of clean food, reach the point where eating more does not automatically mean gaining weight, and do it gradually. He ties that to two outcomes: more potential for muscle growth and better metabolism, which then improves contest prep and preserves more muscle during prep. Whether or not a coach uses the exact same language, the structure is useful. The weekly check-in should answer a specific question: is the athlete moving toward greater food capacity without losing control? If yes, the plan can progress. If no, the check-in should identify whether the limiting factor is intake tolerance, adherence, or a change in body response.

That is why “what happened this week?” is a weak coaching question. “What changed the next decision?” is better.

The corpus gives a clear example with fruit and carbohydrate placement. A client asks whether banana versus blueberries matter on medium days, pre- and post-workout, and whether pasta once a day is fine. Justin’s answer is the kind of answer that improves weekly triage: he expects no noticeable difference over a year from fruit versus other carb sources if macros are handled well, but he still says the last few percent matters. He allows fruit pre- and post-workout up to about 50% of the carbs in those meals with no downside, is fine with fruit in other meals on medium days so long as it is not every meal and at least half the carbs per meal come from more complex sources, and says on high days sugar barely matters because carbs are high and insulin is elevated all day. The decision lesson is not “fruit is magic” or “fruit is irrelevant.” It is that check-ins should separate the high-leverage variables from the fine-tuning variables. Macros and adherence rank above food cosmetics; food cosmetics rank above internet folklore.

That same sorting problem is what AI coaches must learn if they are going to be useful instead of noisy. A weekly check-in can present ten inputs, but only a few should drive action. Appetite, fatigue, body weight trend, food tolerance, insulin response, training performance, and compliance are the obvious candidates. The coach’s job is to decide whether the week is a “hold,” a “nudge,” or a “change.” The bad version of AI coaching overreacts to every note and turns one week into a new model. The good version asks whether the client is still executing the current model cleanly. If yes, leave it alone. If no, adjust the smallest piece that fixes the bottleneck.

Justin’s style in the voice exemplars is consistent with that. He is blunt about tradeoffs. With growth hormone and worsening blood sugar, he does not pretend the issue is abstract: GH can win for bodybuilding while losing on the blood sugar front, and if reduced sensitivity makes the current dose too much, he would drop it, watch blood glucose over a month or two, and possibly hold off for prep. That is not a claim that every problem has an obvious answer. It is a claim that a weekly check-in should produce a decision with a time frame attached. You do not just describe the problem. You assign the observation window and the next move.

For coaches using AI, that is the real standard.

If the check-in creates more words but no better ranking of problems, the system is failing. If it tells you that a high-day insulin dose now needs to come down, that appetite suppression is too strong for a gaining phase, or that a small nutrition detail can wait because macros are already on track, the system is doing its job. The point is not to replace judgment. The point is to make judgment faster, cleaner, and more consistent across weeks.

The practical rule is easy to test: every weekly check-in should end with one of three outputs — hold, adjust, or monitor — and a reason attached to the one variable that actually changed the decision. If AI cannot do that, it is generating commentary, not coaching.

Sources Used

• 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/rory_lazowski___members-c5balaovjbdoeefqmfuqdhh2tbpmfdu16lnf0tnrtmw.md
• modules/03-knowledge/kahunas-coaching-deep-nutrition.md
• modules/08-voice/kahunas-coaching-deep-voice.md
• raw/_consumed/2026-05-26/troponiniq_kb.md