The 7-Minute Check-In Triage Rule

The 7-Minute Check-In Triage Rule

AI coaching gets useful when it routes weekly check-ins by decision quality, not by volume of data.

The strongest pattern in the Kahunas coaching corpus is simple: Justin Harris keeps making the same kind of call from different data streams — reduce the dose, hold the change, or wait for another check-in — when appetite, blood glucose, or food tolerance move in a predictable direction. The mechanism is a feedback loop, not a prediction engine. That matters because AI fitness coaching will not be judged by how much it can summarize; it will be judged by whether it improves weekly triage. My thesis is sharp and falsifiable: if an AI coach cannot sort a check-in into the right next action faster and more consistently than a competent human coach, it is not yet coaching — it is just reporting.

Check-ins are not logs; they are triage packets

Most coaches already know this in practice. The problem is that software often treats a weekly check-in like a scrapbook: bodyweight, steps, mood, photos, training notes, maybe a glucose trend if the athlete uses one. Useful? Sometimes. Decision-ready? Not automatically.

Justin’s actual decision pattern in the KB is more disciplined than the usual “analyze everything” habit. When a lifter’s insulin sensitivity improved and the same dose started driving blood sugar down earlier than expected on a high day, the response was not philosophical. The call was to reduce the dose further on the next high day and bring the meals closer together as needed. That is triage: identify the variable that changed, decide whether the current setup still fits, and make the smallest change that keeps the plan functioning.

The same logic shows up in the retatrutide exchange. Justin reports that he tried it himself, noticed appetite suppression, and was unsure about the idea of forcing appetite lower. But the decision point was not “is this interesting?” It was whether appetite suppression helps or harms the current phase. His provisional call was to run with it while leaning out, then use that period as a better example of how it behaves before deciding what it means for gaining. Again: phase-specific triage, not abstract debate.

Weekly check-in triage should ask one question first: what changed the decision?

A good weekly check-in is not a full history. It is a change detector.

That means the first task is to identify the thing that changed relative to last week. In the Joe Webb exchange, the change was not “insulin is bad” or “the plan is broken.” The change was that the same insulin dose now dipped blood sugar more noticeably after a high day. Once that change is identified, the options narrow fast. Lower the dose, bring meals closer together, and observe the next high day.

AI can help here if it is trained to triage around deltas rather than averages. A check-in that shows stable bodyweight, slightly better training performance, and a new appetite crash is not a “mixed report” in the abstract. It is a decision with one dominant variable: the appetite shift may need to be interpreted as a phase constraint rather than a motivation problem.

That is the useful role for AI coaching systems like TroponinIQ: not to replace judgment with a score, but to help a coach classify the check-in into one of a few action buckets:

• keep the plan unchanged
• reduce the dose or load that is clearly overreaching
• pause the new variable and re-test
• hold steady until the next signal arrives

Those buckets are boring on purpose. Boring is good when the goal is better decisions.

The best triage is conservative, because false certainty is expensive

The KB examples reward restraint. Justin does not oversell effects that are uncertain, and he does not force a change just because a variable is novel.

In the retatrutide exchange, he explicitly says he is holding his opinion until he gets more data. That is not indecision. It is the correct response when a new tool has a strong acute effect — less appetite, more fatigue — but the long-term use case is not yet clear. Coaches get into trouble when they treat early signal as final truth. AI systems do this too, especially when they are optimized to sound helpful every time.

The same restraint appears in the nutrition guidance around carb sources. On the one hand, the KB says the vast majority of results come from nailing macros, and the differences among carb choices are small. On the other hand, the last few percent lives in the details, and fruit may be limited in certain meal contexts depending on the day type. That is a mature coaching pattern: broad constraints first, details second, novelty last.

Translated into weekly check-in triage, the order of operations is:

1. Is the macro outcome still on track?
2. Is there a clear phase-specific problem?
3. Is the new issue big enough to change the plan now?
4. If not, wait and re-check.

If an AI coach cannot preserve that sequence, it will overreact to noise.

Decision quality beats completeness

This is where many fitness tech products miss the mark. They add more inputs, more charts, more “insights,” and assume that more context equals better coaching. But weekly check-in triage is a decision-quality problem, not a data-collection problem.

A strong coach can look at a short report and make a high-quality call because the report is organized around leverage points. Is appetite down sharply? Is blood glucose responding differently to the same dose? Did the athlete change food timing, source, or day structure? Those are decision inputs.

A weak system produces an encyclopedia of observations and leaves the coach to perform the triage manually. That’s wasted attention.

If AI is going to matter in coaching, it should do three things reliably:

1) Detect what is new

Not what is impressive, not what is noisy, just what changed since last check-in.

2) Map the change to a likely mechanism

Appetite suppression, altered insulin sensitivity, food volume tolerance, phase mismatch, or a dose that is now too aggressive.

3) Recommend the smallest reversible action

Reduce, hold, or pause — not because those are magical words, but because they are the safest way to test causality in a living training plan.

That framework is visible in the KB sources. The coach does not chase every fluctuation. He responds when the change crosses a threshold that affects execution.

What this means for AI fitness coaching right now

The near-term win for AI is not glamorous. It is weekly check-in triage that helps a coach answer one question: what do we do next?

If the athlete’s food tolerance is improving, the plan may continue. If the same insulin dose is now causing earlier drops, the plan changes. If a new appetite suppressant is making intake too easy to undercut in the gaining phase, the variable gets re-evaluated. If the data are interesting but not decision-changing, the answer is to wait.

That is the standard coaches should demand from AI: not a summary, but a routing decision with a reason attached.

A practical test is whether the system can produce the same next-step recommendations a good coach would make from the same weekly check-in, without pretending uncertainty is certainty. If it can’t, it may still be useful as a note-taker or organizer. But it has not earned the label “coach.”

TroponinIQ’s value proposition should be judged on that standard. Can it help surface the one or two changes that actually matter? Can it preserve the coach’s conservative sequence of action? Can it keep the weekly check-in focused on reversible decisions instead of performance theater?

That is the point of triage: not to know everything, but to choose the next move well.

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
• modules/03-knowledge/kahunas-coaching-deep-nutrition.md
• modules/08-voice/kahunas-coaching-deep-voice.md