Weekly Check-Ins and the 1-IU Insulin Triage Rule

Weekly Check-Ins and the 1-IU Insulin Triage Rule

AI coaching is only as good as the weekly decision it helps you make: adjust the smallest lever that explains the data, then wait for the next check-in.

Justin Harris’s coaching note on a high day is blunt: when the same insulin dose started dipping blood sugar earlier than usual, he told the client to reduce the dose further on the next high day. That’s the mechanism in miniature: sensitivity drift. The falsifiable thesis is simple — weekly check-ins are not for collecting more data, they are for making one clean triage decision from the smallest signal that actually changed, and better coaching means fewer unnecessary changes, not more.

The job of a check-in is triage, not theater

Most coaches say they want “better adherence” or “better feedback.” That is too vague to be useful. In practice, a weekly check-in has one real purpose: decide whether the current plan still matches the athlete’s current state. If it does, leave it alone. If it doesn’t, identify the narrowest lever that explains the mismatch and change only that.

The Rory Lazowski exchange shows the same logic from a different angle. Justin tried retatrutide himself, reported that it lowered appetite and increased fatigue, and then immediately questioned whether it belonged in a phase where food was about to go up. He did not treat “appetite suppression” as automatically good. He treated it as context-dependent. If the athlete is in a phase where the goal is to push food upward, a drug that makes eating harder is no longer a neutral add-on; it becomes a planning variable. That is triage.

This matters because AI coaching often fails in the exact spot where human coaches get sloppy too: it can record every check-in detail and still miss the decision. The useful question is not “what happened this week?” It is “what changed enough to justify a change?”

Start with the sign that moved

Joe Webb’s note is a clean example. He said his insulin sensitivity seemed improved because the same insulin dose as the prior week dipped blood sugar noticeably earlier than usual, forcing him to eat meal 2 about 30 minutes sooner. He did not overeat. He did not panic. He adjusted the dose on that high day and planned to reduce it further next time.

That is exactly the kind of response quality a check-in system should reward:

• same dose
• different response
• timing changed, not just feelings
• smallest plausible correction

The coach’s task is to keep the change local to the problem. If blood sugar is dropping earlier, the decision is not “rewrite the whole diet.” It is “reduce the dose further on the next high day and see if the pattern resolves.” That protects decision quality because it preserves interpretability. If you change three things at once, you learn nothing.

This is where AI can be helpful and where it can also be dangerous. Helpful, if it surfaces the one variable that moved. Dangerous, if it turns every check-in into a kitchen sink of interpretations. Coaches do not need more commentary; they need cleaner triage.

The best check-ins collapse ambiguity

Justin’s voice across the KB is consistent: he likes decisive calls when the signal is clear and refuses to pretend uncertainty is wisdom. In the GH example from the deep voice module, the issue is not “is GH magical?” It is “what happens when a tool that helps physique outcomes worsens blood sugar?” His answer is a tradeoff, not a slogan: if the only question is bodybuilding, GH wins; if the question is health, it is working against the target. He notes a response curve with diminishing returns and says higher year-round doses push most people down the bad side of that curve.

That same pattern applies to weekly check-ins at a much smaller scale. Good triage asks:

1. What target are we actually optimizing?
2. Which variable moved first?
3. What is the smallest change that addresses it?
4. What do we expect to see next week if we were right?

If the plan cannot answer those four questions, it is not a check-in system; it is a conversation.

Why weekly cadence beats emotional overreaction

Weekly check-ins work because they are slow enough to avoid noise and fast enough to catch drift. Most physique clients do not need minute-by-minute course correction. They need enough time for a change to show up, then enough discipline to avoid overreacting to it.

That is why the Joe Webb example is useful beyond insulin. The client noticed a pattern, described it in terms of timing and dose, and communicated the consequence without dramatizing it. That gives the coach a decision surface, not a pile of feelings. If the coach sees the same pattern in food volume, hunger, bodyweight, training performance, or recovery, the same rule should apply: identify the variable with the clearest causal link and adjust it first.

This is also why the retatrutide discussion matters. Justin’s first concern was not “new tool, new excitement.” It was whether the tool conflicts with the next phase of the plan. If appetite is already low, or if food needs to come up, then appetite suppression is not an abstract benefit. It is a constraint. Weekly triage should catch that conflict before the coach accidentally builds the next block on top of the wrong premise.

What good AI should do in a check-in

If an AI coaching layer is actually useful, it should not try to sound wise. It should do three boring things very well:

• flag the highest-signal change
• map that change to the smallest likely lever
• force a decision that can be tested next week

That is a practical standard, and it is falsifiable. If the AI keeps producing broader and broader plans while the athlete’s problem is becoming more specific, it is failing. If it keeps suggesting changes without stating what outcome should move by next check-in, it is failing. If it can’t tell the difference between a transient fluctuation and a genuine drift in the plan, it is failing.

The KB examples point to a simple coaching principle: the best weekly check-ins are not exhaustive. They are selective. They do not reward the longest explanation. They reward the clearest triage.

The decision quality test

A coach can use one simple test on every weekly check-in: after reading the update, can you name the one thing you would change, the one thing you would not change, and the one result you expect by next week?

If yes, you probably have a real coaching decision. If no, you probably have noise dressed up as attention.

That is the part AI should sharpen. Not more chatter. Better triage.

Sources Used

• raw/_consumed/2026-05-26/troponiniq_kb.md
• raw/Justin_TT1.txt
• modules/03-knowledge/kahunas-coaching-deep-nutrition.md
• modules/08-voice/kahunas-coaching-deep-voice.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/rory_lazowski___members-c5balaovjbdoeefqmfuqdhh2tbpmfdu16lnf0tnrtmw.md