What the AI found
“Your lowest HRV scores consistently appear on Tuesdays, particularly after Monday interval training sessions, averaging 18ms lower than your weekly mean, suggesting over-reaching early in the week.”
Before
Disparate daily metrics, no clear recovery plan
After
Integrated weekly recovery insights, data-driven adjustments
The same system, three states — real screens, not a screenshot
| Date | Sleep Duration — HRV — Training Load (TSS) — Subjective Feeling |
| 2024-03-20 | 7h 12m — 62ms — 150 — Good |
| 2024-03-21 | 6h 45m — 58ms — 210 — Tired |
| 2024-03-22 | 7h 05m — 65ms — 90 — Okay |
Prompt
Here is my last month of daily recovery data, including sleep duration, HRV, training stress score (TSS) from my Garmin, and a subjective feeling score (1=poor, 5=excellent). Can you identify any recurring patterns or correlations, especially regarding days with significantly lower HRV or higher subjective fatigue, and suggest potential triggers based on the training load or specific day of the week?
Here is my last month of daily recovery data, including sleep duration, HRV, training stress score (TSS) from my Garmin, and a subjective feeling score (1=poor, 5=excellent). Can you identify any recurring patterns or correlations, especially regarding days with significantly lower HRV or higher subjective fatigue, and suggest potential triggers based on the training load or specific day of the week?
AI
Reviewing your data, your lowest HRV scores consistently appear on Tuesdays, particularly after Monday interval training sessions. On these days, your average HRV is 48ms, which is 18ms lower than your weekly mean of 66ms. This pattern suggests that your current Monday training intensity might be impacting your early-week recovery more than previously thought.-18 ms
Avg. Weekly Monday HRV Drop
90% on Tuesdays
Consistency of HRV Dip (Avg.)
-15%
Recommended Monday TSS Adj.
A Cyclist’s New View on Recovery
How a customised AI system converted scattered daily metrics into a clear, actionable weekly recovery strategy.
A 41-year-old amateur cyclist, EU
Tools used
The real tools used here — swap any for your own equivalent. Each links to how we’d set it up.
Starting state
Before anything was set up
Our cyclist, a keen amateur based in the EU, diligently tracked a range of metrics: sleep duration and quality via Oura, training load from her Garmin watch, and daily subjective well-being in a simple spreadsheet. The intention was good, but the daily numbers felt like noise. There was no clear way to connect the dots, let alone derive actionable insights for recovery.
| Date | Sleep Duration — HRV — Training Load (TSS) — Subjective Feeling |
| 2024-03-20 | 7h 12m — 62ms — 150 — Good |
| 2024-03-21 | 6h 45m — 58ms — 210 — Tired |
| 2024-03-22 | 7h 05m — 65ms — 90 — Okay |
| ... |
Working state
Setup, doing its job
To cut through this noise, she used Setup to build a simple AI-powered review. Her primary goal: understand the patterns in her recovery data. She extracted a month of data from Oura, Garmin Connect, and her subjective well-being spreadsheet, consolidating it into a Google Sheet. Pasting this data directly into the chat, she then asked for specific trends and correlations with her HRV.
Prompt
Here is my last month of daily recovery data, including sleep duration, HRV, training stress score (TSS) from my Garmin, and a subjective feeling score (1=poor, 5=excellent). Can you identify any recurring patterns or correlations, especially regarding days with significantly lower HRV or higher subjective fatigue, and suggest potential triggers based on the training load or specific day of the week?
Here is my last month of daily recovery data, including sleep duration, HRV, training stress score (TSS) from my Garmin, and a subjective feeling score (1=poor, 5=excellent). Can you identify any recurring patterns or correlations, especially regarding days with significantly lower HRV or higher subjective fatigue, and suggest potential triggers based on the training load or specific day of the week?
AI
Reviewing your data, your lowest HRV scores consistently appear on Tuesdays, particularly after Monday interval training sessions. On these days, your average HRV is 48ms, which is 18ms lower than your weekly mean of 66ms. This pattern suggests that your current Monday training intensity might be impacting your early-week recovery more than previously thought.Use case implemented
The finished system, running on its own
The implemented system now runs weekly. Every Sunday morning, she exports the latest week’s data into her Google Sheet, which is then fed into the AI. She receives a concise summary of recovery trends and specific recommendations for the upcoming training week. This systematic approach ensures her training intensity aligns with her body's actual recovery status, preventing overtraining.
-18 ms
Avg. Weekly Monday HRV Drop
90% on Tuesdays
Consistency of HRV Dip (Avg.)
-15%
Recommended Monday TSS Adj.
What an outside observer would notice
↓ 25%
HRV standard deviation
↑ 15%
Subjective fatigue scores
↑ 30%
Adherence to recovery periods
The stack — build it yourself
Reliable, passive daily capture of crucial recovery markers.
Accurate, integrated platform for recording and quantifying training stress.
Flexible, widely accessible for consolidating data from disparate sources.
Powerful natural language processing for pattern detection and actionable summaries.
These are the tools used in this story. Any can be swapped for an equivalent you already trust.
Go deeper
Do this yourself
See Setup
This story runs on Setup. The tools and prompts above are the real build — swap any tool for your own equivalent and follow the same steps.