Weekly Check-In Triage: 3 Decisions That Improve AI Coaching

Decision quality beats more data when the check-in has to turn into action.

The 2024-12-02 Rory Lazowski exchange shows the mechanism most coaches miss: appetite suppression and fatigue can change the next week’s plan before the spreadsheet does. In that case, Justin Harris did not chase the “helpful in gaining” idea, and he did not freeze the plan because the client had a strong first impression on 2 mg retatrutide; he treated it as a triage problem, noted the appetite drop, noted the fatigue, and said to keep running it while leaning out a bit because body comp was moving in the right direction. That is the core test for weekly AI coaching check-ins: the best system is not the one that collects the most notes, but the one that converts the right notes into a better decision, faster, at the moment the plan changes.

The real job of the weekly check-in

Most coaching check-ins are framed as reporting. That is the wrong mental model. A useful check-in is a triage queue. It asks three questions in order:

Did the body respond as expected?
Did the plan create a new problem?
Does the next week need an adjustment, a hold, or no change?

If you answer those in order, decision quality goes up. If you answer them out of order, coaches get dragged into noise: vanity metrics, emotional overreaction, and changes made just because something feels different.

Justin’s actual coaching style in the KB is consistent on this point. In the Joe Webb check-in, the client noticed that the same insulin dose on a high day was dipping blood sugar more than the prior week, enough that he had to bring meals closer together. Justin did not turn that into a dramatic intervention. He treated the signal as a dose-timing issue, and the client already had the right next step in motion: reduce the dose further on the next high day. That is what good triage looks like. A changed response is not automatically a failed plan; it is a cue to narrow the adjustment to the part that moved.

Decision quality starts with the smallest useful change

AI coaching gets worse when it tries to sound decisive without being specific. The practical rule is simpler: make the smallest change that fits the signal.

That pattern appears across the KB. In the Rory exchange, Justin’s reaction to retatrutide was not “everyone should use this” or “never use this.” It was more grounded: the appetite suppression was real, fatigue was real, and the move was to continue while leaning out because the current phase could use the appetite help. He also said he didn’t like the idea of forcing appetite lower, which matters because it keeps the coach from turning a client-specific tool into a universal answer.

For weekly check-ins, that means the AI should separate three types of response:

Expected response: the plan is working as intended.
Constraint response: the plan works, but it now creates friction.
Mismatch response: the plan is pushing against the goal.

That distinction matters more than polish. A client who is getting hungrier on a higher food target, less hungry on a new compound, or more glucose-sensitive on a high day may all be “responding,” but they are not responding in the same way. A check-in that blurs those together creates bad changes.

What to ask first in a weekly check-in

If the goal is better decisions, the first questions should be about function, not sentiment.

Start with:

Did appetite, fatigue, digestion, or training output change?
Did the response improve, worsen, or just shift timing?
Is the change helping the current phase, or making it harder to execute?

That order matters because it prevents overreacting to data that is real but not important.

Justin’s off-season nutrition comments in the podcast source point in the same direction. He frames the offseason as teaching the body to digest and assimilate a massive amount of clean food, not simply adding muscle by force. The mechanism is digestive and metabolic capacity: the better the body handles food, the better the prep and the better the muscle retention later. That is not just a philosophical point; it tells you what the check-in should be tracking. If a client’s weekly report shows better food tolerance, stable weight at higher intake, and smoother execution, that is useful. If it shows a number changed but the client’s digestion, hunger, and meal compliance are still intact, the week may be a pass.

When to change the plan, and when to leave it alone

Good weekly triage makes fewer changes than bad weekly triage. That sounds conservative, but it is actually aggressive in the right way: it protects the few variables that matter.

Use this sequence:

1. Hold when the signal is expected and the client is executing well.

If the body is responding as planned, changing the plan early usually adds noise. Many AI coaching systems are too eager to “optimize” because they confuse activity with progress. The best move is often to preserve the current dose, food structure, or cardio prescription and watch the next week.

2. Adjust when the signal is real but local.

This is the Joe Webb case. The response changed on the high day, so the fix was not to rewrite the diet; it was to reduce insulin further on the next high day and bring the meals closer together as needed. Local problem, local fix.

3. Reconsider when the response works against the phase.

The Rory retatrutide discussion is the clearest example. If a tool lowers appetite in a phase where the priority is easy leaning out, that can be useful. If the same tool makes gaining harder or flatly suppresses intake when food volume is the growth lever, then the tool is no longer aligned with the objective. Justin explicitly said he was unconvinced by the “helpful in gaining” claim and wanted more data. That is the correct level of skepticism: don’t turn a client’s short-term appetite change into a theory of long-term value.

What AI should do better than a human inbox

Weekly check-ins usually fail because they are unstructured. The coach gets a long paragraph, remembers the loudest problem, and misses the actual decision point. AI should be better at that.

A practical AI triage layer should:

extract the phase of the plan first;
identify whether the signal is appetite, weight trend, training output, blood sugar response, or digestion;
classify the change as expected, constraint, or mismatch;
propose the smallest valid action.

That is useful because it preserves coach judgment where it matters. The AI should not be trying to “decide bodybuilding” in the abstract. It should be narrowing the check-in to the one change that needs a response.

This is also where hype gets people in trouble. A chatbot that produces a confident plan every time is not necessarily helping. A check-in system that reliably says “nothing to change,” “reduce the dose on the high day,” or “this is helping the current phase, continue” will beat a system that tries to be clever.

The falsifiable thesis

If your weekly check-in process cannot separate expected response from local constraint and phase mismatch, AI coaching will add noise faster than it adds value. If it can, it will improve decision quality without needing more data than the coach already had.

That thesis is testable in practice. Look at the next 10 check-ins. Count how often the system recommends a global change when a local fix would have worked, or when a hold would have been enough. If that number is high, the system is failing triage. If it is low, the system is helping the coach think.

Bottom line

Weekly check-ins are not a recap. They are a decision filter. The best AI coaching use case is not writing more commentary; it is sorting the signal so the coach can make one good call: hold, adjust, or reconsider.

That is the difference between busy and useful.

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/joe_webb___members-rksigkykimaxwmo_t4_e8nwvbtc2j0etleutkyysads.json
raw/Justin_TT1.txt
modules/03-knowledge/kahunas-coaching-deep-nutrition.md
modules/08-voice/kahunas-coaching-deep-voice.md