Recovery Signal Quality: 1 Coach Rule for Fatigue, Food, and Patience

Recovery Signal Quality: 1 Coach Rule for Fatigue, Food, and Patience

When recovery data is noisy, the right next move is usually not more training data or more AI—it’s deciding whether the bottleneck is nutrition, workload, or time.

Justin Harris’s repeated coaching call in the Rory Lazowski log was simple: when appetite drops and fatigue rises after a 2 mg retatrutide dose, he would pause or reduce it if food intake needs to climb. That is a mechanism story, not a vibe story: appetite suppression plus lower energy availability can look like “recovery” progress in a dashboard while actually making training feel flatter. The falsifiable thesis is blunt: in AI-assisted coaching, the highest-value recovery signal is not a score, but whether the next adjustment is food, training, or patience.

Recovery is not one metric; it is a decision tree

Most coaching software is eager to compress recovery into a single number. HRV, sleep, soreness, readiness, wellness scores, morning glucose, wearable strain: all useful in isolation, all misleading when treated as a verdict. The practical question is not “Is the athlete recovered?” It is “What changed, and what should change next?”

Justin’s own coaching language keeps returning to that sequence. In one Rory exchange, after the client reported that retatrutide at 2 mg caused “no appetite whatsoever” and “a bit more fatigued than normal,” Justin did not reach for a mystical recovery explanation. He treated the signal as a tradeoff: appetite lower, fatigue higher, likely reduced intake, therefore be cautious about increasing calories with the drug still in play. That is exactly the kind of decision AI should support and not obscure.

The reason matters. Recovery signals are only useful if they are linked to a plausible mechanism. If appetite is down, food intake tends to fall. If food intake falls, training output can flatten even when the athlete feels “disciplined.” If training output flattens, the coach may wrongly intensify programming when the actual fix is simply to eat. The signal quality problem is not that data are unavailable. It is that the wrong data get promoted above the mechanism.

Appetite loss can masquerade as recovery

A lot of fitness technology rewards visible compliance. Fewer calories, lower scale weight, cleaner macro execution, less hunger on paper: these can all look like progress. But the Rory messages show the opposite risk clearly. Justin’s response to retatrutide was not “great, that means the plan is working.” It was “if we’re going to start adding more food in, I may give the reta a pause or at least reduce the dose until prep.”

That sentence encodes a coach’s hierarchy:

1. Is appetite suppression helping the current phase?
2. Is it interfering with food intake needed for the next phase?
3. If so, do not overinterpret the recovery signal.

That hierarchy is useful beyond drugs or peptides. Any intervention that improves one marker while worsening intake, sleep, or training tolerance creates a false recovery story. AI systems can make this worse if they overweight one noisy metric. A readiness score that looks better because resting heart rate dropped is not a win if the athlete is under-eating, dragging through sessions, or needing to move meals closer together just to get through the day.

The falsifiable coaching rule is this: if appetite is suppressed and fatigue is up, and the phase requires more food or more training density, the next change is usually not “train harder.” It is nutrition first, then dose/structure, then patience.

The cleanest signal is phase-specific

Justin’s off-season comments in the podcast source point to the same principle from the nutrition side. He framed the off-season as teaching the body to “digest and assimilate a massive amount of clean food,” with the goal of being able to eat more without gaining excessive weight. The point was not gluttony; it was capacity. Better digestion and assimilation improve the metabolism available for prep and help preserve more muscle in contest prep.

That matters because recovery is phase dependent. In a gaining phase, a small appetite dip can be a big problem. In a cut, appetite suppression may be useful for a while, but only if it does not compromise training recovery so much that muscle retention suffers. The same signal can mean different things depending on the phase.

This is where coaching AI often breaks. It treats the athlete as if the same threshold applies year-round. But the right question is always contextual: what is the objective now? If the objective is to grow, suppressed appetite is not a neutral change. If the objective is to lean out, suppressed appetite may be strategically helpful until it starts reducing training quality or making recovery too expensive.

What the data can actually tell you

The Joe Webb exchange gives a smaller but very coachable example. After a high day, Joe reported improved insulin sensitivity: the same insulin dose as the previous week caused a noticeable drop in blood sugar, so he had to bring meals closer together and reduce the dose further on the next high day. Justin’s role there was not to celebrate a “better recovery score.” It was to interpret a changed response and adjust the next input.

That is the kind of pattern coaches should want AI to flag:

• same dose, different response;
• same meal timing, different glucose effect;
• same workload, different fatigue report.

Those deltas are actionable. They are more useful than a generic readiness score because they identify whether the bottleneck is fuel handling, training load, or time. In practice, the decision is not abstract:

• If a changed response tracks with lower intake, fix nutrition.
• If a changed response tracks with a workload jump, adjust training.
• If there is no clear trend and the athlete is simply in a bad stretch, do not overreact; wait for more data.

That last point matters more than people admit. Patience is not passivity. It is a choice to avoid changing three variables because one dashboard number moved. In recovery coaching, too many interventions are really anxiety in disguise.

What AI should do better than a spreadsheet

The best use of AI here is not prediction theater. It is summarization with mechanism. A good coach-facing system should answer four questions in plain language:

• What changed from last week?
• Does the change point toward food, training, or patience?
• What evidence supports that call?
• What should not change yet?

That’s it. No mystical “recovery index” pretending to know the athlete better than the coach does.

Justin’s bodybuilder-style nutrition framing also hints at a better model for software. He cares about whether the body can handle more clean food, not just whether the scale moved. He cares about preserving muscle in prep, not just whether calories were hit. And in the retatrutide exchange, he cares about whether appetite suppression is now working against the phase. That is recovery signal quality: not how many metrics you have, but whether they point to the correct next decision.

Practical coaching takeaways

If you are using AI or wearables with athletes, the shortest useful workflow is:

• Nutrition first when appetite, intake, or meal timing has clearly changed.
• Training first when output, tolerance, or soreness tracks with a workload increase.
• Patience first when the signal is small, mixed, or obviously phase-dependent.

And one more rule: if the athlete is getting “better” on paper while feeling flatter in the gym, don’t let the paper win automatically. Recovery is not reduced to a score because the score cannot tell you whether the next move is to eat more, back off, or wait.

The best AI coaching systems will not be the ones that shout the loudest about readiness. They will be the ones that help a coach decide, with discipline, whether the next lever is nutrition, training, or patience.

Sources Used

• raw/_consumed/2026-05-31/kahunas-export/2026-05-31-w13-18m/clients/rory_lazowski___members-c5balaovjbdoeefqmfuqdhh2tbpmfdu16lnf0tnrtmw.json
• 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-w19-24m/clients/rory_lazowski___members-c5balaovjbdoeefqmfuqdhh2tbpmfdu16lnf0tnrtmw.json
• modules/03-knowledge/kahunas-coaching-deep-nutrition.md
• raw/Justin_on_Podcast.txt