Weekly Triage 3: The Check-In Decision Stack

Weekly Triage 3: The Check-In Decision Stack

AI coaching gets useful when it routes noisy weeks into a few stable decisions: keep, adjust, or escalate.

The strongest pattern in the Kahunas export is not that coaches asked for more data; it is that Justin Harris repeatedly interpreted weekly check-ins as a triage system, where the first job is to sort signal from noise before changing the plan. In one thread, a big weight drop and visible leanness were weighed against constipation, distension, travel, and water intake; in another, an athlete was told that at the current body fat and doses, there was “zero concern” about being in shape on time. The mechanism is decision gating: don’t let transient review noise trigger unnecessary plan churn. The falsifiable thesis is simple: the best AI check-in systems should improve triage quality, not increase the number of adjustments.

Why weekly check-ins fail

Most coaching check-ins are overloaded by the wrong kind of information. Bodyweight swings, gut fullness, sleep, travel, stress, and subjective confidence all show up at once. If the system treats every input as equally actionable, the coach ends up reacting to artifacts. That is exactly how bad decisions compound: the plan gets changed because a week looked weird, not because the underlying trend changed.

The Kahunas notes give a practical contrast. One message points to a weight drop and a leaner look, but it also flags constipation and distension, with the coach reasoning that travel, lower water intake, and prep-related gut slowdown were more likely drivers than a true physique problem. The conclusion was not “panic and overhaul.” It was “tighten the interpretation.” That is what triage is supposed to do.

In plain terms, a good weekly check-in answers three questions in order:

1. Is the athlete on trend?
2. Is there a plausible non-programming explanation for the noisy part?
3. Does the noise threaten the next decision window?

If the answer to 1 is yes and 2 is yes, the correct move is often to hold steady.

The decision stack coaches actually need

AI coaching tools are often sold as if they should generate more insight. That is the wrong KPI. Coaches do not need a dozen extra interpretations. They need a consistent hierarchy for deciding what matters.

A useful check-in stack looks like this:

Tier 1: outcome trend

• Bodyweight trend over several check-ins
• Visual trend from photos
• Performance trend in the main training tasks

Tier 2: confounders

• Travel
• Water intake changes
• GI upset or constipation
• Sleep disruption
• Unusual soreness or illness-like fatigue

Tier 3: action urgency

• Keep the current plan
• Make a small local adjustment
• Escalate because the trend is genuinely off

That order matters. If you begin with the confounder, you risk over-explaining the week. If you begin with the outcome trend, you preserve the signal.

The message about being backed up after drinking slightly less water on a drive back from KC is a useful example. The coach did not turn a short-lived GI issue into a nutrition crisis. Instead, the issue was contextualized as a sensitivity problem that could recur under similar conditions. That is the level of practical reasoning AI should support: not diagnosis, just clean sorting.

What the AI should do first

If you are building or using AI coaching, the first function should be classification, not recommendation.

A check-in triage model should label each report into something like:

• Stable / stay course
• Noisy but explainable / monitor
• Trend break / intervene

That sounds simple, but simplicity is the point. Coaches lose decision quality when every message gets treated like a fire drill.

The best AI helper is the one that can say:

• “Bodyweight is down, visuals are better, and the gut issue has a plausible cause. Hold.”
• “Performance is flat, but travel and poor intake explain the week. Don’t rewrite the plan yet.”
• “The trend is worsening over multiple checks with no good confounder. Escalate.”

Notice what is missing: drama. The tool does not need to be persuasive. It needs to be consistent.

What the examples imply about weekly review quality

The David LaMartina thread shows how a coach can notice a real issue without overreacting to it. The weight drop was framed positively, but the distension was not ignored. Instead, the likely cause was narrowed down to travel and prep-related constipation. That kind of interpretation keeps coaches from making the common mistake of treating every bad-looking check-in as a plan failure.

The Skip Hill thread adds a second layer. The coach said he did not like prepping hard for longer than 16 weeks, wanted to avoid pushing too much, and had “zero concern” about being in shape on time at the athlete’s current body fat and the doses involved. Whether or not you agree with that programming stance, the decision logic is clear: when the athlete is already in a favorable position, don’t spend the next week chasing changes that are not necessary.

That is the core lesson for AI triage. Better tools do not just report data. They protect the coach from false urgency.

Decision quality beats dashboard density

It is tempting to think the answer is more metrics. More readiness scores. More sentiment labels. More graphs. But the real problem is not lack of information; it is lack of prioritization.

Weekly check-ins should be judged by a boring standard:

• Did they surface the main trend?
• Did they identify the likely confounder?
• Did they preserve the next decision?

If the answer is yes, the check-in was useful.

If AI systems want to earn a place in coaching workflows, they should help with the judgment call that happens after the athlete uploads the pictures and writes the comment. The system should reduce the odds of two bad outcomes:

1. ignoring a real trend because the week looked noisy
2. changing the plan because the week looked noisy

That is a narrow target, but it is the right one.

A practical rule for coaches

When a check-in arrives, force it through this sequence:

1. What is the trend?
2. What is the simplest explanation for the noise?
3. Does this require a plan change this week?

If the answer to step 3 is no, the correct decision is often to document, monitor, and move on.

That may sound underwhelming. It is not. It is what keeps good prep weeks from being ruined by reactive coaching.

The future of AI fitness coaching is not a chatbot that talks louder than the coach. It is a triage layer that helps the coach make fewer bad calls on Mondays.

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/2026-05-31/kahunas-export/2026-05-31-w13-18m/clients/skip_hill___members-b7s-_d0oqqz4a1tadlvuztb0y_vzfqlr9c29tpakweg.json
• raw/_consumed/2026-05-26/troponiniq_kb.md