Check-In Triage: 3 Signals From Weekly Coaching Decisions

Check-In Triage: 3 Signals From Weekly Coaching Decisions

When AI coaching is useful, it is not as a motivational layer but as a sorting layer: it should separate signal from noise, flag the few things that change the plan, and ignore the rest.

The strongest coaching lesson in the corpus is not that more data improves outcomes; it is that conversation sequencing and check-in handling is one of the top topics in the active corpus, which means the bottleneck is decision quality, not raw information volume. In a real client thread, Justin Harris ties a big weight drop to constipation and distension, notes less water on a drive back from KC, and immediately links the problem to a simple mechanism—backup and dehydration in prep—rather than to panic or plan churn. That is the right frame for weekly check-ins: triage the few items that change the program, ignore the noise that does not, and use AI to rank issues before a coach overreacts.

The weekly check-in is not a report; it is a sorting problem

Most weekly check-ins contain more words than decisions. A client may mention bodyweight, sleep, hunger, pumps, digestion, training performance, steps, travel, and stress. The mistake is treating each item as equally important. The better model is triage: decide which signals are actionable now, which need watching, and which are symptoms of a transient context.

That matters because the body does not always change for the reasons a client thinks it changed. In the David LaMartina thread, the coach sees a weight drop and gut distress, but he does not default to “push harder” or “pull food immediately.” He looks at context: travel, less water, constipation, distension. That sequence matters. The decision is not made from the scale alone; it is made from the scale plus the mechanism that plausibly explains it.

That is exactly where AI coaching can earn its keep. Not by replacing the coach’s judgment, and not by generating a glossy summary, but by helping the coach classify the update:

• Red flags: things that require a direct plan change.
• Yellow flags: things worth monitoring next check-in.
• Noise: complaints that are real to the athlete but not currently decision-driving.

If the system cannot do that, it is just text compression.

Decision quality comes from ranking, not from reacting

The best weekly check-in process has a hierarchy. Start with outcomes, then mechanisms, then interventions.

1. Outcome first: What changed? Scale trend, visual trend, performance trend, adherence trend.
2. Mechanism second: Why did it change? Travel, water intake, digestion, stress, schedule disruption, training fatigue.
3. Intervention third: What should the plan actually do next?

This order prevents the classic mistake of confusing a symptom for the cause. Distension is a symptom. Constipation is a mechanism candidate. Less water is a context variable. A coach who reacts to distension by randomly adjusting calories, steps, or cardio is often treating the wrong layer.

The corpus gives a blunt example. Justin’s comment about clients who abuse narcotics getting the worst distention because constipation is one of the main side effects is not a prescription; it is an observational mechanism check. The useful part is the reasoning structure: when gut issues show up, look upstream first. The point is not the specific drug reference. The point is that check-in triage should ask, “What is the likeliest mechanism behind the visible problem?” before changing the macro plan.

AI should pre-sort the check-in before the coach reads it

The most practical use of AI in coaching is to reduce the cognitive load of the weekly review. A strong system can parse a check-in into buckets:

• Body comp trend: weight, photos, look, waist, fullness.
• Recovery trend: sleep, soreness, fatigue, motivation.
• Digestion trend: bowel regularity, bloating, distension, appetite.
• Execution trend: missed meals, water, sodium, schedule changes, travel.
• Training trend: performance, pumps, recovery between sessions.

Then it should highlight only what changed materially since last week.

That sounds obvious, but in practice most “AI coaching” products do the opposite: they summarize everything and therefore prioritize nothing. The coach still has to reread the whole thing and do the triage manually. That is not leverage.

A better workflow is simple:

• AI extracts the deltas.
• AI flags contradictions.
• AI names likely mechanism categories.
• Coach decides whether the plan changes.

If the athlete says “I’m leaner everywhere” and also reports constipation, distension, and lower water intake during travel, the system should not produce equal-weight bullets. It should elevate the travel/water/digestion cluster and downgrade the emotional language. That is decision support.

Good triage prevents bad overcorrection

Weekly check-ins often fail because coaches overcorrect from a single noisy input. One bad night of sleep becomes a training overhaul. One flat look becomes a carb bump. One distended abdomen becomes a full diet rewrite.

The corpus points in the other direction. Justin’s handling of the LaMartina note is practical: he identifies the issue, links it to travel and water, and suggests a simple next-step interpretation. No theatrics. No cascade of unnecessary changes. That is what good triage looks like in coaching: preserve the current plan unless the evidence says the plan is the problem.

This is especially important in prep, where many variables move together. A client can look worse for reasons that have nothing to do with body-fat gain: less water, slower digestion, travel, harder training, more fatigue, less food volume tolerance. If the check-in system cannot separate those, it will misread the athlete’s condition and keep changing levers that did not need changing.

The check-in decision rule coaches can actually use

A useful weekly decision rule is this:

Change the plan only when the same signal appears in two domains or when one domain is extreme.

Examples of domains:

• Body comp
• Performance
• Recovery
• Digestion
• Adherence

If bodyweight is down but performance, look, and adherence are stable, that is not automatically a program problem. If distension appears with travel and reduced water, that is a context problem until proven otherwise. If performance drops, recovery worsens, and adherence slips together, that is a more credible case for intervention.

That rule is falsifiable. It can be tested against the archive of weekly check-ins. It also fits the way experienced coaches already think, which is why AI should support it rather than replace it.

What this means for TroponinIQ-style coaching

TroponinIQ’s value is not that it can produce more words about a check-in. Its value is that it can help a coach make faster, cleaner calls on what matters this week. In other words, the product should be judged by triage quality.

A coach does not need an AI that remembers every sentence. A coach needs an AI that says:

• this is probably digestion, not fat gain;
• this is probably travel-related, not a new program flaw;
• this is the one variable that changed enough to justify action;
• the rest can wait one more week.

That is the practical standard.

If AI coaching cannot improve weekly check-in triage, it is not solving the real bottleneck. If it can, the win is not flashy. It is fewer bad adjustments, cleaner decisions, and better weekly judgment.

Bottom line

Weekly check-in triage is the highest-value use case for AI coaching because the central problem is not information scarcity; it is decision ranking. The evidence in the corpus points to a coach who reads context before changing the plan, and a platform that earns its keep by surfacing mechanism, not noise. The falsifiable thesis is simple: the best AI coaching systems will improve coaching by making weekly check-ins more selective, not more elaborate.

Sources Used

• raw/_consumed/2026-05-31/kahunas-export/2026-05-31-w19-24m/clients/david_lamartina___members-tlssnsjthkmnhfqcscszce25acz_vhdm_x2_xdlpx_i.json
• raw/_consumed/2026-05-26/troponiniq_kb.md
• ingestion-report.md